int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
ESL_RANDOMNESS *r = NULL; /* random number generator */
FILE *ofp = NULL; /* data output stream */
int outfmt = eslSQFILE_FASTA;
/* Parse command line */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK)
cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK)
cmdline_failure(argv[0], "Error in app configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") )
cmdline_help(argv[0], go);
/* Open the output data file, if any */
if (esl_opt_GetString(go, "-o") != NULL)
{
if ((ofp = fopen(esl_opt_GetString(go, "-o"), "w")) == NULL)
esl_fatal("Failed to open output file %s\n", esl_opt_GetString(go, "-o"));
}
else ofp = stdout;
/* Initialize */
r = esl_randomness_Create(esl_opt_GetInteger(go, "--seed"));
/* Hand off execution to one of the three modes */
if (esl_opt_GetBoolean(go, "-A")) /* Alignment shuffling */
{
if (esl_opt_ArgNumber(go) != 1)
cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
msa_shuffling(go, r, ofp, outfmt);
}
else if (esl_opt_GetBoolean(go, "-G")) /* Sequence generation */
{
if (esl_opt_ArgNumber(go) != 0)
cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
seq_generation(go, r, ofp, outfmt);
}
else if (esl_opt_GetBoolean(go, "-S")) /* Sequence shuffling */
{
if (esl_opt_ArgNumber(go) != 1)
cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
seq_shuffling(go, r, ofp, outfmt);
}
if (esl_opt_GetString(go, "-o") != NULL) fclose(ofp);
esl_randomness_Destroy(r);
esl_getopts_Destroy(go);
return 0;
}
/* Function: p7_CreateDefaultApp()
* Synopsis: Initialize a small/simple/standard HMMER application
* Incept: SRE, Thu Oct 28 15:03:21 2010 [Janelia]
*
* Purpose: Identical to <esl_getopts_CreateDefaultApp()>, but
* specialized for HMMER. See documentation in
* <easel/esl_getopts.c>.
*
* Args: options - array of <ESL_OPTIONS> structures for getopts
* nargs - number of cmd line arguments expected (excl. of cmdname)
* argc - <argc> from main()
* argv - <argv> from main()
* banner - optional one-line description of program (or NULL)
* usage - optional one-line usage hint (or NULL)
*
* Returns: ptr to new <ESL_GETOPTS> object.
*
* On command line errors, this routine prints an error
* message to <stderr> then calls <exit(1)> to halt
* execution with abnormal (1) status.
*
* If the standard <-h> option is seen, the routine prints
* the help page (using the data in the <options> structure),
* then calls <exit(0)> to exit with normal (0) status.
*
* Xref: J7/3
*
* Note: The only difference between this and esl_getopts_CreateDefaultApp()
* is to call p7_banner() instead of esl_banner(), to get HMMER
* versioning info into the header. There ought to be a better way
* (perhaps using PACKAGE_* define's instead of HMMER_* vs. EASEL_*
* define's in esl_banner(), thus removing the need for p7_banner).
*/
ESL_GETOPTS *
p7_CreateDefaultApp(ESL_OPTIONS *options, int nargs, int argc, char **argv, char *banner, char *usage)
{
ESL_GETOPTS *go = NULL;
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
if (usage != NULL) esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
if (banner != NULL) p7_banner(stdout, argv[0], banner);
if (usage != NULL) esl_usage (stdout, argv[0], usage);
puts("\nOptions:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != nargs)
{
puts("Incorrect number of command line arguments.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
return go;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
ESL_RANDOMNESS *r = NULL;
int be_verbose;
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK) esl_fatal("%s", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
esl_usage(stdout, argv[0], usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
exit(0);
}
if (esl_opt_ArgNumber(go) != 0) {
printf("Incorrect number of command line arguments.\n");
esl_usage(stdout, argv[0], usage);
exit(1);
}
be_verbose = esl_opt_GetBoolean(go, "-v");
if (esl_opt_GetBoolean(go, "-r")) {
r = esl_randomness_CreateTimeseeded();
if (be_verbose) printf("seed = %ld\n", esl_randomness_GetSeed(r));
} else r = esl_randomness_Create(esl_opt_GetInteger(go, "-s"));
utest_LinearRegression(r, TRUE, be_verbose);
utest_LinearRegression(r, FALSE, be_verbose);
esl_getopts_Destroy(go);
esl_randomness_Destroy(r);
exit(0);
}
static int
process_commandline(int argc, char **argv, ESL_GETOPTS **ret_go, char **ret_fmfile, char **ret_qfile)
{
ESL_GETOPTS *go = esl_getopts_Create(options);
int status;
if (esl_opt_ProcessEnvironment(go) != eslOK) { if (printf("Failed to process environment: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) { if (printf("Failed to parse command line: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_VerifyConfig(go) != eslOK) { if (printf("Failed to parse command line: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* help format: */
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
esl_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
if (puts("\nBasic options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1= group; 2 = indentation; 120=textwidth*/
if (puts("\nSpecial options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80); /* 2= group; 2 = indentation; 120=textwidth*/
exit(0);
}
if (esl_opt_ArgNumber(go) != 2) { if (puts("Incorrect number of command line arguments.") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_qfile = esl_opt_GetArg(go, 1)) == NULL) { if (puts("Failed to get <qfile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_fmfile = esl_opt_GetArg(go, 2)) == NULL) { if (puts("Failed to get <fmfile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* Validate any attempted use of stdin streams */
if (esl_strcmp(*ret_fmfile, "-") == 0 && esl_strcmp(*ret_qfile, "-") == 0)
{ if (puts("Either <fmfile> or <qfile> may be '-' (to read from stdin), but not both.") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
*ret_go = go;
return eslOK;
FAILURE: /* all errors handled here are user errors, so be polite. */
esl_usage(stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1= group; 2 = indentation; 80=textwidth*/
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
esl_getopts_Destroy(go);
exit(1);
ERROR:
if (go) esl_getopts_Destroy(go);
exit(status);
}
static void
process_commandline(int argc, char **argv, ESL_GETOPTS **ret_go, char **ret_hmmfile, char **ret_alifile)
{
ESL_GETOPTS *go = NULL;
if ((go = esl_getopts_Create(options)) == NULL) p7_Die("problem with options structure");
if (esl_opt_ProcessEnvironment(go) != eslOK) { printf("Failed to process environment: %s\n", go->errbuf); goto ERROR; }
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) { printf("Failed to parse command line: %s\n", go->errbuf); goto ERROR; }
if (esl_opt_VerifyConfig(go) != eslOK) { printf("Failed to parse command line: %s\n", go->errbuf); goto ERROR; }
/* help format: */
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
puts("\nOptions for selecting alphabet rather than guessing it:");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
puts("\nAlternative model construction strategies:");
esl_opt_DisplayHelp(stdout, go, 3, 2, 80);
puts("\nAlternative relative sequence weighting strategies:");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
puts("\nAlternate effective sequence weighting strategies:");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
puts("\nControl of E-value calibration:");
esl_opt_DisplayHelp(stdout, go, 6, 2, 80);
puts("\nOther options:");
esl_opt_DisplayHelp(stdout, go, 8, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != 2) { puts("Incorrect number of command line arguments."); goto ERROR; }
if ((*ret_hmmfile = esl_opt_GetArg(go, 1)) == NULL) { puts("Failed to get <hmmfile> argument on command line"); goto ERROR; }
if ((*ret_alifile = esl_opt_GetArg(go, 2)) == NULL) { puts("Failed to get <alifile> argument on command line"); goto ERROR; }
*ret_go = go;
return;
ERROR: /* all errors handled here are user errors, so be polite. */
esl_usage(stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
printf("\nTo see more help on other available options, do %s -h\n\n", argv[0]);
exit(1);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
char *keyfile = NULL;
char *tabfile = NULL;
ESL_KEYHASH *kh = esl_keyhash_Create();
int nkeys = 0;
ESL_DMATRIX *D = NULL;
ESL_TREE *T = NULL;
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], go, "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], go, "Error in app configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) cmdline_help (argv[0], go);
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], go, "Incorrect number of command line arguments.\n");
keyfile = esl_opt_GetArg(go, 1);
tabfile = esl_opt_GetArg(go, 2);
read_keyfile(go, keyfile, kh);
nkeys = esl_keyhash_GetNumber(kh);
D = esl_dmatrix_Create(nkeys, nkeys);
read_tabfile(go, tabfile, kh, D);
esl_tree_SingleLinkage(D, &T);
//esl_tree_WriteNewick(stdout, T);
output_clusters(go, T, kh);
esl_tree_Destroy(T);
esl_dmatrix_Destroy(D);
esl_keyhash_Destroy(kh);
esl_getopts_Destroy(go);
return 0;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go; /* application configuration */
int kstatus, tstatus;/* return code from Easel routine */
int fmt; /* expected format of kfile, tfile */
char *kfile, *tfile; /* known, test structure file */
ESL_MSAFILE *kfp, *tfp; /* open kfile, tfile */
ESL_MSA *ka, *ta; /* known, trusted alignment */
int64_t klen, tlen; /* lengths of dealigned seqs */
int i; /* counter over sequences */
int apos; /* counter over alignment columns */
int rfpos; /* counter over consensus (non-gap RF) columns */
int is_rfpos; /* TRUE if current apos is a consensus pos, FALSE if not */
int uapos; /* counter over unaligned residue positions */
int nali; /* number of alignment we're on in each file */
int **kp; /* [0..i..nseq-1][1..r..sq->n] = x known non-gap RF position of residue r in sequence i */
int **tp; /* [0..i..nseq-1][1..r..sq->n] = x predicted non-gap RF position of residue r in sequence i */
/* for both kp and pp, if x <= 0, residue r for seq i is not aligned to a non-gap RF position, but rather as an 'insert'
* after non-gap RF position (x * -1)
*/
int *km_pos; /* [0..rflen] = x, in known aln, number of residues aligned to non-gap RF column x; special case: mct[0] = 0 */
int *ki_pos; /* [0..rflen] = x, in known aln, number of residues inserted after non-gap RF column x */
int *tm_pos; /* [0..rflen] = x, in predicted aln, number of residues aligned to non-gap RF column x; special case: mct[0] = 0 */
int *ti_pos; /* [0..rflen] = x, in predicted aln, number of residues inserted after non-gap RF column x */
int *cor_tm_pos; /* [0..rflen] = x, in predicted aln, number of correctly predicted residues aligned to non-gap RF column x; special case: mct[0] = 0 */
int *cor_ti_pos; /* [0..rflen] = x, in predicted aln, number of correctly predicted residues inserted after non-gap RF column x */
int *km_seq; /* [0..i..nseq-1] = x, in known aln, number of residues aligned to non-gap RF columns in seq i; */
int *ki_seq; /* [0..i..nseq-1] = x, in known aln, number of residues inserted in seq i */
int *tm_seq; /* [0..i..nseq-1] = x, in predicted aln, number of residues aligned to non-gap RF columns in seq i; */
int *ti_seq; /* [0..i..nseq-1] = x, in predicted aln, number of residues inserted in seq i */
int *cor_tm_seq; /* [0..i..nseq-1] = x, in predicted aln, number of correctly predicted residues aligned to non-gap RF columns in seq i */
int *cor_ti_seq; /* [0..i..nseq-1] = x, in predicted aln, number of correctly predicted residues inserted in seq i */
int *seqlen; /* [0..i..nseq-1] = x, unaligned seq i has length x */
ESL_ALPHABET *abc = NULL; /* alphabet for all alignments */
int rflen, t_rflen; /* non-gap RF length (consensus lengths) */
int status;
char *namedashes;
int ni;
int namewidth = 8; /* length of 'seq name' */
int cor_tm, cor_ti, km, ki; /* correct predicted match, correct predicted insert, total match, total insert */
char *mask = NULL;
int masklen;
ESL_DSQ *ks;
ESL_DSQ *ts;
FILE *dfp = NULL; /* for --c2dfile */
/* variables needed for -p and related options */
int do_post = FALSE; /* TRUE if -p enabled */
int do_post_for_this_rfpos = FALSE; /* set for each consensus position, always TRUE unless --mask-p2xm */
int p; /* counter over integerized posteriors */
int *ptm = NULL; /* [0..p..10] number of total matches with posterior value p (10="*")*/
int *pti = NULL; /* [0..p..10] number of total inserts with posterior value p */
int *cor_ptm = NULL; /* [0..p..10] number of correct matches with posterior value p */
int *cor_pti = NULL; /* [0..p..10] number of correct inserts with posterior value p */
int npostvals = 11; /* number of posterior values 0-9, * */
int ppidx; /* index of PP */
char ppchars[11] = "0123456789*";
int cm_cor_ptm, cm_cor_pti, cm_ptm, cm_pti, cm_incor_ptm, cm_incor_pti; /* cumulative counts of posteriors */
// int tot_cor_ptm, tot_cor_pti, tot_ptm, tot_pti; /* total counts of posteriors */
// int tot_incor_ptm,tot_incor_pti; // SRE: commented out; don't seem to be used; need to silence compiler warning
char errbuf[eslERRBUFSIZE];
/***********************************************
* Parse command line
***********************************************/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") )
{
esl_banner(stdout, argv[0], banner);
esl_usage (stdout, argv[0], usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
exit(EXIT_SUCCESS);
}
if (esl_opt_ArgNumber(go) != 2)
{
printf("Incorrect number of command line arguments.\n");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
kfile = esl_opt_GetArg(go, 1);
tfile = esl_opt_GetArg(go, 2);
fmt = eslMSAFILE_STOCKHOLM;
/***********************************************
* Open the two Stockholm files.
***********************************************/
if (esl_opt_GetBoolean(go, "--amino")) abc = esl_alphabet_Create(eslAMINO);
else if (esl_opt_GetBoolean(go, "--dna")) abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) abc = esl_alphabet_Create(eslRNA);
if ( (kstatus = esl_msafile_Open(&abc, kfile, NULL, fmt, NULL, &kfp)) != eslOK) esl_msafile_OpenFailure(kfp, kstatus);
if ( (tstatus = esl_msafile_Open(&abc, tfile, NULL, fmt, NULL, &tfp)) != eslOK) esl_msafile_OpenFailure(tfp, tstatus);
do_post = esl_opt_GetBoolean(go, "-p");
/* read the mask file if --p-mask is enabled */
if(! esl_opt_IsDefault(go, "--p-mask")) {
if((status = read_mask_file(esl_opt_GetString(go, "--p-mask"), errbuf, &mask, &masklen)) != eslOK) esl_fatal(errbuf);
}
/* open the c2dfile for output, if nec */
if (esl_opt_IsOn(go, "--c2dfile")) {
if ((dfp = fopen(esl_opt_GetString(go, "--c2dfile"), "w")) == NULL) esl_fatal("Failed to open --c2dfile output file %s\n", esl_opt_GetString(go, "--c2dfile"));
}
/***********************************************
* Do alignment comparisons, one seq at a time;
* this means looping over all seqs in all alignments.
***********************************************/
nali = 0;
while ( (kstatus = esl_msafile_Read(kfp, &ka)) != eslEOF)
{
if ( kstatus != eslOK) esl_msafile_ReadFailure(kfp, kstatus);
if ( (tstatus = esl_msafile_Read(tfp, &ta)) != eslOK) esl_msafile_ReadFailure(tfp, tstatus);
nali++;
if((nali > 1) && (esl_opt_IsOn(go, "--c2dfile"))) esl_fatal("--c2dfile is only meant for msafiles with single alignments");
/* Sanity check on alignment
*/
if (ka->nseq != ta->nseq)
esl_fatal("trusted, test alignments don't have same seq #\n");
if (ka->rf == NULL)
esl_fatal("trusted alignment has no reference annotation\n");
if (ta->rf == NULL)
esl_fatal("test alignment has no reference annotation\n");
/* make sure the sequences are all identical */
ESL_ALLOC(seqlen, sizeof(int) * ka->nseq);
for(i = 0; i < ka->nseq; i++) {
if(strcmp(ka->sqname[i], ta->sqname[i]) != 0) esl_fatal("sequence %d of trusted alignment %s has different name than seq %d of predicted alignment %s\n", (i+1), ka->sqname[i], (i+1), ta->sqname[i]);
ESL_ALLOC(ks, sizeof(ESL_DSQ) * (ka->alen+2));
memcpy(ks, ka->ax[i], (ka->alen+2) * sizeof(ESL_DSQ));
esl_abc_XDealign(ka->abc, ks, ka->ax[i], &klen);
ESL_ALLOC(ts, sizeof(ESL_DSQ) * (ta->alen+2));
memcpy(ts, ta->ax[i], (ta->alen+2) * sizeof(ESL_DSQ));
esl_abc_XDealign(ta->abc, ts, ta->ax[i], &tlen);
if (tlen != klen)
esl_fatal("dealigned sequence mismatch, seq %d, when dealigned, is %d residues in the known alignment, but %d residues in the trusted alignment.", (i+1), klen, tlen);
if (memcmp(ks, ts, sizeof(ESL_DSQ) * klen) != 0)
esl_fatal("dealigned sequence mismatch, seq %d %s, when dealigned, are not identical.", (i+1), ka->sqname[i]);
seqlen[i] = tlen;
free(ks);
free(ts);
}
/* determine non-gap RF length */
rflen = 0;
for(apos = 1; apos <= ka->alen; apos++) {
if((! esl_abc_CIsGap (ka->abc, ka->rf[apos-1])) &&
(! esl_abc_CIsMissing(ka->abc, ka->rf[apos-1]))) rflen++;
}
t_rflen = 0;
for(apos = 1; apos <= ta->alen; apos++) {
if((! esl_abc_CIsGap (ta->abc, ta->rf[apos-1])) &&
(! esl_abc_CIsMissing (ta->abc, ta->rf[apos-1]))) t_rflen++;
}
if(t_rflen != rflen) esl_fatal("Trusted alignment non-gap RF length (%d) != predicted alignment non-gap RF length (%d).\n", rflen, t_rflen);
/* if -p, make sure the test alignment has posterior probabilities, and allocate our counters for correct/incorrect per post value */
if(do_post) {
if(! esl_opt_IsDefault(go, "--p-mask")) {
if(masklen != rflen) {
esl_fatal("Length of mask in %s (%d) not equal to non-gap RF len of alignments (%d)\n", esl_opt_GetString(go, "--p-mask"), masklen, rflen);
}
}
if(ta->pp == NULL) esl_fatal("-p requires \"#=GR PP\" annotation in the test alignment, but none exists");
ESL_ALLOC(ptm, sizeof(int) * npostvals);
ESL_ALLOC(pti, sizeof(int) * npostvals);
ESL_ALLOC(cor_ptm, sizeof(int) * npostvals);
ESL_ALLOC(cor_pti, sizeof(int) * npostvals);
esl_vec_ISet(ptm, npostvals, 0);
esl_vec_ISet(pti, npostvals, 0);
esl_vec_ISet(cor_ptm, npostvals, 0);
esl_vec_ISet(cor_pti, npostvals, 0);
}
/* allocate and initialize our counters */
ESL_ALLOC(kp, sizeof(int *) * ka->nseq);
ESL_ALLOC(tp, sizeof(int *) * ta->nseq);
for(i = 0; i < ka->nseq; i++) {
ESL_ALLOC(kp[i], sizeof(int) * (seqlen[i]+1));
ESL_ALLOC(tp[i], sizeof(int) * (seqlen[i]+1));
esl_vec_ISet(kp[i], seqlen[i]+1, -987654321);
esl_vec_ISet(tp[i], seqlen[i]+1, -987654321);
}
ESL_ALLOC(km_pos, sizeof(int) * (rflen+1));
ESL_ALLOC(ki_pos, sizeof(int) * (rflen+1));
ESL_ALLOC(tm_pos, sizeof(int) * (rflen+1));
ESL_ALLOC(ti_pos, sizeof(int) * (rflen+1));
ESL_ALLOC(cor_tm_pos, sizeof(int) * (rflen+1));
ESL_ALLOC(cor_ti_pos, sizeof(int) * (rflen+1));
esl_vec_ISet(km_pos, rflen+1, 0);
esl_vec_ISet(ki_pos, rflen+1, 0);
esl_vec_ISet(tm_pos, rflen+1, 0);
esl_vec_ISet(ti_pos, rflen+1, 0);
esl_vec_ISet(cor_tm_pos, rflen+1, 0);
esl_vec_ISet(cor_ti_pos, rflen+1, 0);
ESL_ALLOC(km_seq, sizeof(int) * ka->nseq);
ESL_ALLOC(ki_seq, sizeof(int) * ka->nseq);
ESL_ALLOC(tm_seq, sizeof(int) * ka->nseq);
ESL_ALLOC(ti_seq, sizeof(int) * ka->nseq);
ESL_ALLOC(cor_tm_seq, sizeof(int) * ka->nseq);
ESL_ALLOC(cor_ti_seq, sizeof(int) * ka->nseq);
esl_vec_ISet(km_seq, ka->nseq, 0);
esl_vec_ISet(ki_seq, ka->nseq, 0);
esl_vec_ISet(tm_seq, ka->nseq, 0);
esl_vec_ISet(ti_seq, ka->nseq, 0);
esl_vec_ISet(cor_tm_seq, ka->nseq, 0);
esl_vec_ISet(cor_ti_seq, ka->nseq, 0);
/* determine non-gap RF location of each residue in known alignment */
for(i = 0; i < ka->nseq; i++) {
uapos = rfpos = 0;
for(apos = 1; apos <= ka->alen; apos++) {
is_rfpos = FALSE;
if((! esl_abc_CIsGap (ka->abc, ka->rf[apos-1])) &&
(! esl_abc_CIsMissing (ka->abc, ka->rf[apos-1]))) {
rfpos++; is_rfpos = TRUE;
}
if(esl_abc_XIsResidue(ka->abc, ka->ax[i][apos])) {
uapos++;
kp[i][uapos] = (is_rfpos) ? rfpos : (-1 * rfpos);
if(is_rfpos) { km_pos[rfpos]++; km_seq[i]++; }
else { ki_pos[rfpos]++; ki_seq[i]++; }
}
}
}
/* determine non-gap RF location of each residue in predicted alignment */
for(i = 0; i < ta->nseq; i++) {
uapos = rfpos = 0;
for(apos = 1; apos <= ta->alen; apos++) {
is_rfpos = FALSE;
if((! esl_abc_CIsGap (abc, ta->rf[apos-1])) &&
(! esl_abc_CIsMissing (abc, ta->rf[apos-1]))) {
rfpos++; is_rfpos = TRUE;
if(do_post) {
do_post_for_this_rfpos = (mask != NULL && mask[rfpos-1] == '0') ? FALSE : TRUE;
}
}
if(esl_abc_XIsResidue(ta->abc, ta->ax[i][apos])) {
uapos++;
tp[i][uapos] = (is_rfpos) ? rfpos : (-1 * rfpos);
if(do_post) {
if(esl_abc_CIsGap(abc, ta->pp[i][(apos-1)])) esl_fatal("gap PP value for nongap residue: ali: %d seq: %d apos: %d\n", nali, i, apos);
ppidx = get_pp_idx(abc, ta->pp[i][(apos-1)]);
if(ppidx == -1) esl_fatal("unrecognized PP value (%c) for nongap residue: ali: %d seq: %d apos: %d\n", ta->pp[i][(apos-1)], nali, i, apos);
}
if(is_rfpos) {
tm_pos[rfpos]++; tm_seq[i]++;
if(do_post_for_this_rfpos) ptm[ppidx]++;
}
else {
ti_pos[rfpos]++; ti_seq[i]++;
if(do_post) pti[ppidx]++;
}
if(kp[i][uapos] == tp[i][uapos]) { /* correctly predicted this residue */
if(is_rfpos) {
cor_tm_seq[i]++; cor_tm_pos[rfpos]++;
if(do_post_for_this_rfpos) cor_ptm[ppidx]++;
}
else {
cor_ti_seq[i]++; cor_ti_pos[rfpos]++;
if(do_post) cor_pti[ppidx]++;
}
}
}
}
}
if((! (esl_opt_GetBoolean(go, "-c"))) && (! esl_opt_GetBoolean(go, "-p"))) {
/* print per sequence statistics */
/* determine the longest name in msa */
for(ni = 0; ni < ka->nseq; ni++) namewidth = ESL_MAX(namewidth, strlen(ka->sqname[ni]));
ESL_ALLOC(namedashes, sizeof(char) * namewidth+1);
namedashes[namewidth] = '\0';
for(ni = 0; ni < namewidth; ni++) namedashes[ni] = '-';
printf("# %-*s %6s %28s %28s %28s\n", namewidth, "seq name", "len", "match columns", "insert columns", "all columns");
printf("# %-*s %6s %28s %28s %28s\n", namewidth, namedashes, "------", "----------------------------", "----------------------------", "----------------------------");
for(i = 0; i < ta->nseq; i++) {
printf(" %-*s %6d %8d / %8d (%.3f) %8d / %8d (%.3f) %8d / %8d (%.3f)\n", namewidth, ka->sqname[i], seqlen[i],
cor_tm_seq[i], km_seq[i], (km_seq[i] == 0) ? 0. : ((float) cor_tm_seq[i] / (float) km_seq[i]),
cor_ti_seq[i], ki_seq[i], (ki_seq[i] == 0) ? 0. : ((float) cor_ti_seq[i] / (float) ki_seq[i]),
(cor_tm_seq[i] + cor_ti_seq[i]), (km_seq[i] + ki_seq[i]), ((float) (cor_tm_seq[i] + cor_ti_seq[i]) / ((float) km_seq[i] + ki_seq[i])));
}
cor_tm = esl_vec_ISum(cor_tm_seq, ka->nseq);
cor_ti = esl_vec_ISum(cor_ti_seq, ka->nseq);
km = esl_vec_ISum(km_seq, ka->nseq);
ki = esl_vec_ISum(ki_seq, ka->nseq);
printf("# %-*s %6s %28s %28s %28s\n", namewidth, namedashes, "-----", "----------------------------", "----------------------------", "----------------------------");
printf("# %-*s %6s %8d / %8d (%.3f) %8d / %8d (%.3f) %8d / %8d (%.3f)\n",
namewidth, "*all*", "-",
cor_tm, km, ((float) cor_tm / (float) km),
cor_ti, ki, ((float) cor_ti / (float) ki),
(cor_tm+cor_ti), (km+ki), (((float) (cor_tm + cor_ti))/ ((float) (km + ki))));
free(namedashes);
for(i = 0; i < ka->nseq; i++) {
free(kp[i]);
free(tp[i]);
}
}
else if(esl_opt_GetBoolean(go, "-c")) { /* print per column statistics */
printf("# %5s %20s %20s %20s\n", "rfpos", "match", "insert", "both");
printf("# %5s %20s %20s %20s\n", "-----", "--------------------", "--------------------", "--------------------");
for(rfpos = 0; rfpos <= rflen; rfpos++) {
printf(" %5d %4d / %4d (%.3f) %4d / %4d (%.3f) %4d / %4d (%.3f)\n", rfpos,
cor_tm_pos[rfpos], km_pos[rfpos], (km_pos[rfpos] == 0) ? 0. : ((float) cor_tm_pos[rfpos] / (float) km_pos[rfpos]),
cor_ti_pos[rfpos], ki_pos[rfpos], (ki_pos[rfpos] == 0) ? 0. : ((float) cor_ti_pos[rfpos] / (float) ki_pos[rfpos]),
(cor_tm_pos[rfpos] + cor_ti_pos[rfpos]), (km_pos[rfpos] + ki_pos[rfpos]), ((float) (cor_tm_pos[rfpos] + cor_ti_pos[rfpos]) / ((float) km_pos[rfpos] + ki_pos[rfpos])));
}
}
else if(do_post) { /* do posterior output */
if(mask == NULL) {
printf("# %2s %29s %29s\n", "", " match columns ", " insert columns ");
printf("# %2s %29s %29s\n", "", "-----------------------------", "-----------------------------") ;
printf("# %2s %8s %8s %9s %8s %8s %9s\n", "PP", "ncorrect", "ntotal", "fractcor", "ncorrect", "ntotal", "fractcor");
printf("# %2s %8s %8s %9s %8s %8s %9s\n", "--", "--------", "--------", "---------", "--------", "--------", "---------");
}
else {
printf("# %2s %29s %29s\n", "", " match columns within mask ", " insert columns ");
printf("# %2s %29s %29s\n", "", "-----------------------------", "-----------------------------") ;
printf("# %2s %8s %8s %9s %8s %8s %9s\n", "PP", "ncorrect", "ntotal", "fractcor", "ncorrect", "ntotal", "fractcor");
printf("# %2s %8s %8s %9s %8s %8s %9s\n", "--", "--------", "--------", "---------", "--------", "--------", "---------");
}
cm_ptm = cm_pti = cm_cor_ptm = cm_cor_pti = cm_incor_ptm = cm_incor_pti = 0;
//tot_ptm = esl_vec_ISum(ptm, npostvals);
//tot_pti = esl_vec_ISum(pti, npostvals);
//tot_cor_ptm = esl_vec_ISum(cor_ptm, npostvals);
//tot_cor_pti = esl_vec_ISum(cor_pti, npostvals);
//tot_incor_ptm = tot_ptm - tot_cor_ptm;
//tot_incor_pti = tot_pti - tot_cor_pti;
for(p = (npostvals-1); p >= 0; p--) {
cm_cor_ptm += cor_ptm[p];
cm_cor_pti += cor_pti[p];
cm_ptm += ptm[p];
cm_pti += pti[p];
cm_incor_ptm += ptm[p] - cor_ptm[p];
cm_incor_pti += pti[p] - cor_pti[p];
printf(" %2c %8d / %8d (%.5f) %8d / %8d (%.5f)\n",
ppchars[p], cor_ptm[p], ptm[p],
(ptm[p] == 0) ? 0. : (float) cor_ptm[p] / (float) ptm[p],
cor_pti[p], pti[p],
(pti[p] == 0) ? 0. : (float) cor_pti[p] / (float) pti[p]);
}
}
/* handle --c2dfile */
if (dfp != NULL) {
/* match stats, 4 fields, CMYK color values */
for(rfpos = 1; rfpos <= rflen; rfpos++) {
if(km_pos[rfpos] == 0) { /* special case, no known alignment residues, a blank position */
fprintf(dfp, "%.3f %.3f %.3f %.3f\n", 0., 0., 0., 0.);
}
else {
fprintf(dfp, "%.3f %.3f %.3f %.3f\n",
0., /* cyan */
1. - ((float) cor_tm_pos[rfpos] / (float) km_pos[rfpos]), /* magenta, fraction incorrect */
1. - ((float) km_pos[rfpos] / ta->nseq), /* yellow, 1 - fraction of seqs with residue in column */
0.);
}
}
fprintf(dfp, "//\n");
/* insert stats, 4 fields, CMYK color values */
rfpos = 0; /* special case, combine insert posn 0 and 1 together */
if(ki_pos[rfpos] == 0) { /* special case, no known alignment residues, a blank position */
fprintf(dfp, "%.3f %.3f %.3f %.3f\n", 0., 0., 0., 0.);
}
else {
fprintf(dfp, "%.3f %.3f %.3f %.3f\n",
0., /* cyan */
1. - ((float) (cor_ti_pos[0] + cor_ti_pos[1]) / ((float) (ki_pos[0] + ki_pos[1]))), /* magenta, fraction correct */
0.,
0.);
}
/* insert stats posn 2..rflen */
for(rfpos = 2; rfpos <= rflen; rfpos++) {
if(ki_pos[rfpos] == 0) { /* special case, no known alignment residues, a blank position */
fprintf(dfp, "%.3f %.3f %.3f %.3f\n", 0., 0., 0., 0.);
}
else {
fprintf(dfp, "%.3f %.3f %.3f %.3f\n",
0., /* cyan */
1. - ((float) cor_ti_pos[rfpos] / (float) ki_pos[rfpos]), /* magenta, fraction correct */
0.,
0.);
}
}
fprintf(dfp, "//\n");
}
if(ptm != NULL) free(ptm);
if(pti != NULL) free(pti);
if(cor_ptm != NULL) free(cor_ptm);
if(cor_ptm != NULL) free(cor_pti);
free(kp);
free(tp);
free(km_seq);
free(ki_seq);
free(tm_seq);
free(ti_seq);
free(cor_tm_seq);
free(cor_ti_seq);
free(km_pos);
free(ki_pos);
free(tm_pos);
free(ti_pos);
free(cor_tm_pos);
free(cor_ti_pos);
free(seqlen);
esl_msa_Destroy(ka);
esl_msa_Destroy(ta);
}
if(mask != NULL) free(mask);
if(dfp != NULL) {
fclose(dfp);
printf("# Draw file of per-column stats saved to file: %s\n", esl_opt_GetString(go, "--c2dfile"));
}
if(abc) esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
esl_msafile_Close(tfp);
esl_msafile_Close(kfp);
return 0;
ERROR:
return status;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
char *hmmfile = NULL; /* HMM file name */
P7_HMMFILE *hfp = NULL; /* open HMM file */
FILE *ofp = NULL; /* output stream for HMMs */
int status; /* easel/hmmer return code */
/***********************************************
* Parse command line
***********************************************/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) cmdline_help (argv[0], go);
if (esl_opt_ArgNumber(go) < 1) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
/* Open the HMM file. */
hmmfile = esl_opt_GetArg(go, 1);
status = p7_hmmfile_Open(hmmfile, NULL, &hfp);
if (status == eslENOTFOUND) p7_Fail("Failed to open HMM file %s for reading.\n", hmmfile);
else if (status == eslEFORMAT) p7_Fail("File %s does not appear to be in a recognized HMM format.\n", hmmfile);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n", status, hmmfile);
/* Open the output file, if any */
if (esl_opt_GetBoolean(go, "-O"))
{
if ((ofp = fopen(esl_opt_GetArg(go, 2), "w")) == NULL)
esl_fatal("Failed to open output file %s\n", esl_opt_GetArg(go, 2));
}
else if (esl_opt_GetString(go, "-o") != NULL)
{
if ((ofp = fopen(esl_opt_GetString(go, "-o"), "w")) == NULL)
esl_fatal("Failed to open output file %s\n", esl_opt_GetString(go, "-o"));
}
else ofp = stdout;
/* Hand off to the appropriate routine */
if (esl_opt_GetBoolean(go, "--index"))
{
if (esl_opt_ArgNumber(go) != 1) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
create_ssi_index(go, hfp);
}
else if (esl_opt_GetBoolean(go, "-f"))
{
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
multifetch(go, ofp, esl_opt_GetArg(go, 2), hfp);
}
else
{
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
onefetch(go, ofp, esl_opt_GetArg(go, 2), hfp);
if (ofp != stdout) printf("\n\nRetrieved HMM %s.\n", esl_opt_GetArg(go, 2));
}
if (esl_opt_GetBoolean(go, "-O") || esl_opt_GetString(go, "-o") != NULL) fclose(ofp);
p7_hmmfile_Close(hfp);
esl_getopts_Destroy(go);
exit(0);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line processing */
ESL_ALPHABET *abc = NULL;
char *hmmfile = NULL;
P7_HMMFILE *hfp = NULL;
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
int nhmm;
double x;
float KL;
int status;
char errbuf[eslERRBUFSIZE];
float nseq;
int do_eval2score = 0;
int do_score2eval = 0;
int z_val;
float e_val;
float s_val;
/* Process the command line options.
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
puts("\nOptions:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=docgroup, 2 = indentation; 80=textwidth*/
exit(0);
}
if (esl_opt_ArgNumber(go) != 1)
{
puts("Incorrect number of command line arguments.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if ((hmmfile = esl_opt_GetArg(go, 1)) == NULL)
{
puts("Failed to read <hmmfile> argument from command line.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
output_header(stdout, go);
if ( esl_opt_IsOn(go, "--eval2score") ) {
do_eval2score = TRUE;
e_val = esl_opt_GetReal(go, "-E");
} else if ( esl_opt_IsOn(go, "--score2eval") ) {
do_score2eval = TRUE;
s_val = esl_opt_GetReal(go, "-S");
} else if ( esl_opt_IsUsed(go, "--baseZ") || esl_opt_IsUsed(go, "--baseZ1") || esl_opt_IsUsed(go, "-Z") ) {
puts("The flags -Z, --baseZ, and --baseZ1 are for use with --eval2score and --score2eval.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_IsUsed(go, "--baseZ") ) {
z_val = 1000000 * 2 * (long)(esl_opt_GetInteger(go, "--baseZ"));
} else if (esl_opt_IsUsed(go, "--baseZ1") ) {
z_val = 1000000 * (long)(esl_opt_GetInteger(go, "--baseZ1"));
} else {
z_val = esl_opt_GetInteger(go, "-Z");
}
/* Initializations: open the HMM file
*/
status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf);
/* Main body: read HMMs one at a time, print one line of stats
*/
printf("#\n");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s", "idx", "name", "accession", "nseq", "eff_nseq", "M", "relent", "info", "p relE", "compKL");
if (do_eval2score)
printf (" %6s %6.2g", "sc for", e_val);
if (do_score2eval)
printf (" %6s %6.2f", "E-val for", s_val);
printf("\n");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s", "----", "--------------------", "------------", "--------", "--------", "------", "------", "------", "------", "------");
if (do_eval2score)
printf (" %13s", "-------------");
if (do_score2eval)
printf (" %13s", "-------------");
printf("\n");
nhmm = 0;
while ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) != eslEOF)
{
if (status == eslEOD) esl_fatal("read failed, HMM file %s may be truncated?", hmmfile);
else if (status == eslEFORMAT) esl_fatal("bad file format in HMM file %s", hmmfile);
else if (status == eslEINCOMPAT) esl_fatal("HMM file %s contains different alphabets", hmmfile);
else if (status != eslOK) esl_fatal("Unexpected error in reading HMMs from %s", hmmfile);
nhmm++;
if ( esl_opt_IsOn(go, "--eval2score") || esl_opt_IsOn(go, "--score2eval") ) {
if (esl_opt_IsUsed(go, "--baseZ") || esl_opt_IsUsed(go, "--baseZ1" ) ) {
if ( hmm->abc->type != eslRNA && hmm->abc->type != eslDNA) {
puts("The flags --baseZ and --baseZ1 can't be used with non-nucleotide models.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
} else if ( hmm->abc->type != eslAMINO && hmm->abc->type != eslRNA && hmm->abc->type != eslDNA) {
puts("The flags --eval2score and --score2eval can't be used with non-sequence models.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
}
if (esl_opt_IsUsed(go, "--baseZ") ) {
nseq = (float)z_val / (float)(hmm->max_length);
} else if (esl_opt_IsUsed(go, "--baseZ1") ) {
nseq = (float)z_val / (float)(hmm->max_length);
} else {
nseq = z_val;
}
if (bg == NULL) bg = p7_bg_Create(abc);
p7_MeanPositionRelativeEntropy(hmm, bg, &x);
p7_hmm_CompositionKLDist(hmm, bg, &KL, NULL);
printf("%-6d %-20s %-12s %8d %8.2f %6d %6.2f %6.2f %6.2f %6.2f",
nhmm,
hmm->name,
hmm->acc == NULL ? "-" : hmm->acc,
hmm->nseq,
hmm->eff_nseq,
hmm->M,
p7_MeanMatchRelativeEntropy(hmm, bg),
p7_MeanMatchInfo(hmm, bg),
x,
KL);
if ( do_eval2score )
{
float sc;
sc = esl_exp_invsurv( e_val / nseq , hmm->evparam[p7_FTAU], hmm->evparam[p7_FLAMBDA]);
printf("%13.2f", sc);
}
else if ( do_score2eval)
{
float e;
e = nseq * esl_exp_surv( s_val , hmm->evparam[p7_FTAU], hmm->evparam[p7_FLAMBDA]);
printf("%13.2g", e);
}
printf("\n");
/* p7_MeanForwardScore(hmm, bg)); */
p7_hmm_Destroy(hmm);
}
p7_bg_Destroy(bg);
esl_alphabet_Destroy(abc);
p7_hmmfile_Close(hfp);
esl_getopts_Destroy(go);
exit(0);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go;
char *msafile;
ESLX_MSAFILE *afp;
ESL_MSA *msa;
float *sqd;
int status;
int nbad;
int nali = 0;
int nbadali = 0;
int nwgt = 0;
int nbadwgt = 0;
int i;
int be_quiet;
int do_gsc;
int do_pb;
int do_blosum;
double maxid;
double tol;
int maxN;
/* Process command line
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal("failed to parse cmd line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal("failed to parse cmd line: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
puts(usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
return 0;
}
be_quiet = esl_opt_GetBoolean(go, "-q");
do_blosum = esl_opt_GetBoolean(go, "--blosum");
do_gsc = esl_opt_GetBoolean(go, "--gsc");
do_pb = esl_opt_GetBoolean(go, "--pb");
maxid = esl_opt_GetReal (go, "--id");
tol = esl_opt_GetReal (go, "--tol");
maxN = esl_opt_GetInteger(go, "--maxN");
if (esl_opt_ArgNumber(go) != 1) {
puts("Incorrect number of command line arguments.");
puts(usage);
return 1;
}
msafile = esl_opt_GetArg(go, 1);
esl_getopts_Destroy(go);
/* Weight one or more alignments from input file
*/
if ((status = eslx_msafile_Open(NULL, msafile, NULL, eslMSAFILE_UNKNOWN, NULL, &afp)) != eslOK)
eslx_msafile_OpenFailure(afp, status);
while ( (status = eslx_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) eslx_msafile_ReadFailure(afp, status);
if (maxN > 0 && msa->nseq > maxN) { esl_msa_Destroy(msa); continue; }
nali++;
nwgt += msa->nseq;
ESL_ALLOC(sqd, sizeof(float) * msa->nseq);
if (do_gsc) {
esl_msaweight_GSC(msa);
GSCWeights(msa->aseq, msa->nseq, msa->alen, sqd);
} else if (do_pb) {
esl_msaweight_PB(msa);
PositionBasedWeights(msa->aseq, msa->nseq, msa->alen, sqd);
} else if (do_blosum) {
esl_msaweight_BLOSUM(msa, maxid);
BlosumWeights(msa->aseq, msa->nseq, msa->alen, maxid, sqd);
/* workaround SQUID bug: BLOSUM weights weren't renormalized to sum to nseq. */
esl_vec_FNorm (sqd, msa->nseq);
esl_vec_FScale(sqd, msa->nseq, (float) msa->nseq);
}
if (! be_quiet) {
for (i = 0; i < msa->nseq; i++)
fprintf(stdout, "%-20s %.3f %.3f\n",
msa->sqname[i], msa->wgt[i], sqd[i]);
}
nbad = 0;
for (i = 0; i < msa->nseq; i++)
if (esl_DCompare((double) sqd[i], msa->wgt[i], tol) != eslOK)
nbad++;
if (nbad > 0) nbadali++;
nbadwgt += nbad;
if (nbad > 0) printf("%-20s :: alignment shows %d weights that differ (out of %d) \n",
msa->name, nbad, msa->nseq);
esl_msa_Destroy(msa);
free(sqd);
}
eslx_msafile_Close(afp);
if (nbadali == 0)
printf("OK: all weights identical between squid and Easel in %d alignment(s)\n", nali);
else {
printf("%d of %d weights mismatched at (> %f fractional difference)\n",
nbadwgt, nwgt, tol);
printf("involving %d of %d total alignments\n", nbadali, nali);
}
return eslOK;
ERROR:
return status;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
ESL_RANDOMNESS *r = NULL;
char **as = NULL; /* aligned character seqs (random, iid) */
int N,L; /* # of seqs, and their aligned lengths */
int seed;
int i,j;
int status;
double p[4]; /* ACGT probabilities */
#ifdef eslAUGMENT_ALPHABET
ESL_DSQ **ax = NULL; /* digitized alignment */
ESL_ALPHABET *abc = NULL;
#endif
/* Process command line
*/
go = esl_getopts_Create(options);
esl_opt_ProcessCmdline(go, argc, argv);
esl_opt_VerifyConfig(go);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
puts(usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
return 0;
}
L = esl_opt_GetInteger(go, "-L");
N = esl_opt_GetInteger(go, "-N");
seed = esl_opt_GetInteger(go, "--seed");
if (esl_opt_ArgNumber(go) != 0) {
puts("Incorrect number of command line arguments.");
puts(usage);
return 1;
}
esl_getopts_Destroy(go);
/* Create a random DNA alignment;
* force it to obey the conventions of the unit tests:
* 0,1 are identical
* 0,2 are completely dissimilar
*/
r = esl_randomness_Create(seed);
for (i = 0; i < 4; i++) p[i] = 0.25;
ESL_ALLOC(as, sizeof(char *) * N);
for (i = 0; i < N; i++)
ESL_ALLOC(as[i], sizeof(char) * (L+1));
esl_rsq_IID(r, "ACGT", p, 4, L, as[0]);
strcpy(as[1], as[0]);
esl_rsq_IID(r, "ACGT", p, 4, L, as[2]);
for (j = 0; j < L; j++)
while (as[2][j] == as[0][j])
as[2][j] = "ACGT"[esl_rnd_Roll(r, 4)];
for (i = 3; i < N; i++)
esl_rsq_IID(r, "ACGT", p, 4, L, as[i]);
#ifdef eslAUGMENT_ALPHABET
abc = esl_alphabet_Create(eslDNA);
ESL_ALLOC(ax, sizeof(ESL_DSQ *) * N);
for (i = 0; i < N; i++)
esl_abc_CreateDsq(abc, as[i], &(ax[i]));
#endif /*eslAUGMENT_ALPHABET*/
/* Unit tests
*/
if (utest_CPairId(as, N) != eslOK) return eslFAIL;
if (utest_CJukesCantor(4, as, N) != eslOK) return eslFAIL;
#ifdef eslAUGMENT_ALPHABET
if (utest_XPairId(abc, as, ax, N) != eslOK) return eslFAIL;
if (utest_XJukesCantor(abc, as, ax, N) != eslOK) return eslFAIL;
#endif /*eslAUGMENT_ALPHABET*/
#ifdef eslAUGMENT_DMATRIX
if (utest_CPairIdMx(as, N) != eslOK) return eslFAIL;
if (utest_CDiffMx(as, N) != eslOK) return eslFAIL;
if (utest_CJukesCantorMx(4, as, N) != eslOK) return eslFAIL;
#endif /* eslAUGMENT_DMATRIX*/
#if defined (eslAUGMENT_ALPHABET) && defined (eslAUGMENT_DMATRIX)
if (utest_XPairIdMx(abc, as, ax, N) != eslOK) return eslFAIL;
if (utest_XDiffMx(abc, as, ax, N) != eslOK) return eslFAIL;
if (utest_XJukesCantorMx(abc, as, ax, N) != eslOK) return eslFAIL;
#endif
esl_randomness_Destroy(r);
esl_Free2D((void **) as, N);
#ifdef eslAUGMENT_ALPHABET
esl_alphabet_Destroy(abc);
esl_Free2D((void **) ax, N);
#endif
return eslOK;
ERROR:
return eslFAIL;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
ESL_ALPHABET *abc = NULL; /* biological alphabet */
char *alifile = NULL; /* alignment file name */
int fmt; /* format code for alifiles */
ESL_MSAFILE *afp = NULL; /* open alignment file */
ESL_MSA *msa = NULL; /* multiple sequence alignment */
int status; /* easel return code */
int do_info = TRUE; /* TRUE if -i */
int do_max = FALSE; /* TRUE if -x */
int do_ffreq = FALSE; /* TRUE if --ffreq */
int do_fmin = FALSE; /* TRUE if --fmin */
float fthresh = 0.; /* <x> from -f <x> */
int do_remove_bps = FALSE; /* TRUE if -r */
int do_consistent = FALSE; /* TRUE if -c */
int do_indi2cons = FALSE; /* TRUE if --indi <x> */
int have_cons; /* TRUE if first alignment has consensus sequence */
int do_newcons = FALSE; /* TRUE if we're creating a new consensus structure
* and outputing a new alignment (if -x -f -c or --indi)
*/
int do_a = FALSE; /* TRUE if -a */
char *indi; /* for <x> from --indi <x> */
int nindi_read; /* number of individual sequence SS lines we've read for current alignment */
int a; /* counter over seqs */
int i, i2; /* counter over residues */
int j, j2; /* counter over residues */
int nali; /* counter over alignments */
int **bp = NULL; /* bp[i][j] is number of individual bps exist between aln cols i and j */
int *cur_ct = NULL; /* ct array of basepairs for current sequence */
int *cons_ct = NULL; /* ct array of basepairs for SS_cons being created */
int *xcons_ct = NULL; /* ct array of basepairs for existing SS_cons */
int *ngaps = NULL; /* number of gaps in each alignment position */
FILE *ofp; /* output file (default is stdout) */
int be_verbose = FALSE; /* TRUE to print extra info */
int seqthresh; /* sequence number threshold for defining a bp as consensus (int) ((fthresh * nseq) + 0.5)*/
char *sscons = NULL; /* the new SS_cons line */
FILE *lfp = NULL; /* file to list sequences with conflicting bps to */
int nlist = 0; /* number of sequences listed to list file */
int *nconflictsA; /* number of conflicting bps in seq a's individual structure annotation */
int nconflicts_total = 0; /* total number of conflicts */
int nconflicts_list = 0; /* total number of conflicts in sequences listed to file <x> from -l <x> */
int noverlaps_total = 0; /* total number of overlaps */
int nconsistent_total = 0; /* total number of consistent bps */
int nbps_total = 0; /* total number of bps */
int *nconsistentA; /* number of consistent bps in seq a's individual structure annotation */
int *noverlapsA; /* number of bps in seq a's indi structure that overlap with consensus structure */
int *nbpsA; /* number of bps in seq a's indi structure that overlap with consensus structure */
int ncons_bps = 0; /* number of bps in consensus structure */
int max_noverlaps_aidx;
int max_nconsistent_aidx;
int max_nbps_aidx;
int *removebp; /* removebp[i] is TRUE remove consensus bp [i]:xcons_ct[i] */
int *has_conflict;
int *nmates_l2r; /* half matrix, nmate_l2r[i] = <x>, i < nmate_l2r[i], there are <x> different right mates j for i */
int *nmates_r2l; /* half matrix, nmate_r2l[j] = <x>, j < nmate_r2l[j], there are <x> different left mates i for j */
int lmax; /* with -l, maximum number of conflicts to allow */
int namewidth = 18; /* length of 'SS_cons(consensus)' */
char *namedashes = NULL; /* to store underline for seq name */
/* --fmin related variables */
int nbps = 0;
int prev_nbps = -1;
float fmin;
int inconsistent_flag;
int pknot_flag;
int k,l;
/***********************************************
* Parse command line
***********************************************/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") )
{
esl_banner(stdout, argv[0], banner);
esl_usage (stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
puts("\noptions for defining a new consensus structure (all of these require -o):");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
puts("\noptions for listing sequences based on structure:");
esl_opt_DisplayHelp(stdout, go, 3, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != 1)
{
printf("Incorrect number of command line arguments.\n");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
alifile = esl_opt_GetArg(go, 1);
fmt = eslMSAFILE_STOCKHOLM;
/***********************************************
* Open the MSA file; determine alphabet; set for digital input
***********************************************/
if (esl_opt_GetBoolean(go, "--dna")) abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) abc = esl_alphabet_Create(eslRNA);
if ( (status = esl_msafile_Open(&abc, alifile, NULL, fmt, NULL, &afp)) != eslOK)
esl_msafile_OpenFailure(afp, status);
/* open output file */
if (esl_opt_GetString(go, "-o") != NULL) {
if ((ofp = fopen(esl_opt_GetString(go, "-o"), "w")) == NULL)
esl_fatal("Failed to open -o output file %s\n", esl_opt_GetString(go, "-o"));
} else ofp = NULL;
if (esl_opt_GetString(go, "-l") != NULL) {
if ((lfp = fopen(esl_opt_GetString(go, "-l"), "w")) == NULL)
esl_fatal("Failed to open -l output file %s\n", esl_opt_GetString(go, "-l"));
}
/* determine if we're creating a structure */
do_max = esl_opt_GetBoolean(go, "-x");
if(!(esl_opt_IsDefault(go, "--ffreq"))) {
do_ffreq = TRUE;
fthresh = esl_opt_GetReal(go, "--ffreq");
}
if(!(esl_opt_IsDefault(go, "--fmin"))) {
do_fmin = TRUE;
}
do_remove_bps = esl_opt_GetBoolean(go, "-r");
do_consistent = esl_opt_GetBoolean(go, "-c");
if(!(esl_opt_IsDefault(go, "--indi"))) {
do_indi2cons = TRUE;
}
if(do_max || do_ffreq || do_fmin || do_remove_bps || do_consistent || do_indi2cons) {
do_newcons = TRUE;
}
do_a = esl_opt_GetBoolean(go, "-a");
if(do_a || do_max || do_ffreq || do_fmin || do_remove_bps || do_consistent || do_indi2cons) {
do_info = FALSE;
}
/***********************************************
* Read MSAs one at a time.
***********************************************/
nali = 0;
have_cons = FALSE;
lmax = esl_opt_GetInteger(go, "--lmax");
if(esl_opt_GetBoolean(go, "-v")) be_verbose = TRUE;
while ((status = esl_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) esl_msafile_ReadFailure(afp, status);
nali++;
/* determine max length name */
namewidth = 18; /* length of 'SS_cons(consensus)' */
for(i = 0; i < msa->nseq; i++) namewidth = ESL_MAX(namewidth, strlen(msa->sqname[i]));
if(namedashes != NULL) { free(namedashes); }
ESL_ALLOC(namedashes, sizeof(char) * namewidth+1);
namedashes[namewidth] = '\0';
for(i = 0; i < namewidth; i++) namedashes[i] = '-';
ESL_ALLOC(sscons, sizeof(char) * (msa->alen+1));
ESL_ALLOC(cur_ct, sizeof(int) * (msa->alen+1));
ESL_ALLOC(cons_ct, sizeof(int) * (msa->alen+1));
ESL_ALLOC(xcons_ct, sizeof(int) * (msa->alen+1));
ESL_ALLOC(bp, sizeof(int *) * (msa->alen+1));
ESL_ALLOC(removebp, sizeof(int) * (msa->alen+1));
ESL_ALLOC(has_conflict, sizeof(int) * (msa->alen+1));
ESL_ALLOC(nmates_l2r, sizeof(int) * (msa->alen+1));
ESL_ALLOC(nmates_r2l, sizeof(int) * (msa->alen+1));
esl_vec_ISet(cur_ct, (msa->alen+1), 0);
esl_vec_ISet(cons_ct, (msa->alen+1), 0);
esl_vec_ISet(xcons_ct, (msa->alen+1), 0);
esl_vec_ISet(removebp, (msa->alen+1), FALSE);
esl_vec_ISet(has_conflict, (msa->alen+1), FALSE);
esl_vec_ISet(nmates_l2r, (msa->alen+1), 0);
esl_vec_ISet(nmates_r2l, (msa->alen+1), 0);
ESL_ALLOC(nconflictsA, sizeof(int) * msa->nseq);
ESL_ALLOC(noverlapsA, sizeof(int) * msa->nseq);
ESL_ALLOC(nconsistentA, sizeof(int) * msa->nseq);
ESL_ALLOC(nbpsA, sizeof(int) * msa->nseq);
esl_vec_ISet(nconflictsA, msa->nseq, 0);
esl_vec_ISet(noverlapsA, msa->nseq, 0);
esl_vec_ISet(nconsistentA, msa->nseq, 0);
esl_vec_ISet(nbpsA, msa->nseq, 0);
max_noverlaps_aidx = max_nconsistent_aidx = max_nbps_aidx = 0;
nconsistent_total = nbps_total = noverlaps_total = nconflicts_total = nconflicts_list = 0;
for(i = 1; i <= msa->alen; i++) {
ESL_ALLOC(bp[i], sizeof(int) * (msa->alen+1));
esl_vec_ISet(bp[i], (msa->alen+1), 0);
}
/* make sure we have ss_cons and indi ss if we need it */
if(msa->ss_cons == NULL && do_remove_bps) esl_fatal("-r requires all alignments have SS_cons annotation, alignment %d does not.", nali);
if(msa->ss == NULL && do_max) esl_fatal("-x requires all alignments have individual SS annotation, alignment %d does not.", nali);
if(msa->ss == NULL && do_consistent) esl_fatal("-c requires all alignments have individual SS annotation, alignment %d does not.", nali);
if(msa->ss == NULL && do_indi2cons) esl_fatal("--indi requires all alignments have individual SS annotation, alignment %d does not.", nali);
if(msa->ss == NULL && do_ffreq) esl_fatal("--ffreq requires all alignments have individual SS annotation, alignment %d does not.", nali);
if(msa->ss == NULL && do_fmin) esl_fatal("--fmin requires all alignments have individual SS annotation, alignment %d does not.", nali);
if(msa->ss_cons != NULL) {
if((status = esl_wuss2ct(msa->ss_cons, msa->alen, xcons_ct)) != eslOK) {
esl_fatal("Existing SS_cons for alignment %d is invalid.", nali);
}
ncons_bps = 0;
for(i = 1; i <= msa->alen; i++)
if(xcons_ct[i] != 0 && i < xcons_ct[i])
ncons_bps++;
if(nali > 1 && !have_cons)
esl_fatal("the first aln has SS_cons but aln %d lacks it, if one has it, they all must.", nali);
if(nali == 1) have_cons = TRUE;
}
else if (lfp != NULL) {
esl_fatal("the -l option requires existing SS_cons annotation, aln %d lacks it.", nali);
}
else if (do_remove_bps) {
esl_fatal("the -r option requires existing SS_cons annotation, aln %d lacks it.", nali);
}
else if (do_consistent) {
esl_fatal("the -c option requires existing SS_cons annotation, aln %d lacks it.", nali);
}
else {
if(nali > 1 && have_cons)
esl_fatal("the first aln does not have SS_cons but aln %d does, if one has it, they all must.", nali);
}
if(do_info) {
printf("# Per-sequence basepair information:\n");
printf("# Alignment file: %s\n", alifile);
printf("# Alignment idx: %d\n", nali);
if(msa->name != NULL) { printf("# Alignment name: %s\n", msa->name); }
if(have_cons) {
printf("#\n");
printf("# indibp: number of basepairs in the individual sequence SS annotation\n");
printf("# ovrlap: number of indibp basepairs that also exist as consensus basepairs\n");
printf("# cnsist: number of indibp basepairs that do not conflict with any consensus basepairs\n");
printf("# cnflct: number of indibp basepairs that conflict with >= 1 consensus basepairs\n");
printf("#\n");
printf("# A conflict exists between two basepairs in different structures, one between columns i and j\n");
printf("# and the other between columns k and l, if (i == k and j != l) or (j == l and i != k).\n");
printf("#\n");
printf("# %-*s %6s %6s %6s %6s\n", namewidth, "seqname", "indibp", "ovrlap", "cnsist", "cnflct");
printf("# %-*s %6s %6s %6s %6s\n", namewidth, namedashes, "------", "------", "-----", "------");
}
else {
printf("# %-*s %6s\n", namewidth, "seqname", "nbp");
printf("# %-*s %6s\n", namewidth, namedashes, "------");
}
}
nindi_read = 0;
for (a = 0; a < msa->nseq; a++) {
if(msa->ss != NULL && msa->ss[a] != NULL) {
if((status = esl_wuss2ct(msa->ss[a], msa->alen, cur_ct)) != eslOK) {
esl_fatal("SS annotation for sequence %d, aln %d is invalid.\n", (a+1), nali);
}
nindi_read++;
for(i = 1; i <= msa->alen; i++) {
if(i < cur_ct[i]) {
bp[i][cur_ct[i]]++;
if(bp[i][cur_ct[i]] == 1) {
nmates_l2r[i]++;
nmates_r2l[cur_ct[i]]++;
}
}
}
for(i = 1; i <= msa->alen; i++) {
if(cur_ct[i] != 0 && i < cur_ct[i]) {
if(xcons_ct[i] == cur_ct[i]) noverlapsA[a]++;
if((xcons_ct[i] != 0) && (xcons_ct[i] != cur_ct[i])) {
if(be_verbose) { printf("ali: %2d seq %3d (%s) bp %4d:%4d conflicts with consensus bp %4d:%4d\n", nali, a, msa->sqname[a], i, cur_ct[i], i, xcons_ct[i]); }
nconflictsA[a]++;
/* indi bp i:cur_ct[i] conflicts with i:xcons_ct[i] */
removebp[i] = TRUE;
removebp[xcons_ct[i]] = TRUE;
}
else if((xcons_ct[cur_ct[i]] != 0) && (xcons_ct[cur_ct[i]] != i) && (cur_ct[xcons_ct[cur_ct[i]]] == 0)) {
if(be_verbose) { printf("ali: %2d seq %3d (%s) bp %4d:%4d conflicts with consensus bp %4d:%4d\n", nali, a, msa->sqname[a], xcons_ct[i], cur_ct[xcons_ct[i]], xcons_ct[cur_ct[i]], cur_ct[i]); }
nconflictsA[a]++;
/* indi bp i:cur_ct[i] conflicts with xcons_ct[cur_ct[i]]:cur_ct[i] */
removebp[cur_ct[i]] = TRUE;
removebp[xcons_ct[cur_ct[i]]] = TRUE;
}
else nconsistentA[a]++;
}
}
if(nconflictsA[a] > lmax) {
if(lfp != NULL) fprintf(lfp, "%s\n", msa->sqname[a]);
nconflicts_list += nconflictsA[a];
nlist++;
}
nbpsA[a] = nconflictsA[a] + nconsistentA[a];
nconflicts_total += nconflictsA[a];
nconsistent_total += nconsistentA[a];
noverlaps_total += noverlapsA[a];
nbps_total += nbpsA[a];
if(do_info && have_cons) printf(" %-*s %6d %6d %6d %6d\n", namewidth, msa->sqname[a], nbpsA[a], noverlapsA[a], nconsistentA[a], nconflictsA[a]);
if(do_info && !have_cons) printf(" %-*s %6d\n", namewidth, msa->sqname[a], nbpsA[a]);
if(nbpsA[a] > nbpsA[max_nbps_aidx]) max_nbps_aidx = a;
if((noverlapsA[a] > noverlapsA[max_noverlaps_aidx]) || ((noverlapsA[a] == noverlapsA[max_noverlaps_aidx]) && (nbpsA[a] > nbpsA[max_noverlaps_aidx]))) max_noverlaps_aidx = a;
if((nconsistentA[a] > nconsistentA[max_nconsistent_aidx]) || ((nconsistentA[a] == nconsistentA[max_nconsistent_aidx]) && (nbpsA[a] > nbpsA[max_nconsistent_aidx]))) max_nconsistent_aidx = a;
}
else if(do_newcons || esl_opt_GetBoolean(go, "-a")) { esl_fatal("No SS annotation for sequence %d, aln %d.\n", (a+1), nali); }
}
if(do_info && have_cons) {
if(nindi_read > 0) printf("\n");
printf(" %-*s %6d %6d %6d %6d\n", namewidth, "SS_cons(consensus)", ncons_bps, ncons_bps, ncons_bps, 0);
if(nindi_read > 0) {
printf("\n# %6d/%6d (%.3f) overlap\n", noverlaps_total, nbps_total, nbps_total > 0 ? (float) noverlaps_total / (float) nbps_total : 0.);
printf("# %6d/%6d (%.3f) consistent\n", nconsistent_total, nbps_total, nbps_total > 0 ? (float) nconsistent_total / (float) nbps_total: 0.);
printf("# %6d/%6d (%.3f) conflict\n", nconflicts_total, nbps_total, nbps_total > 0 ? (float) nconflicts_total / (float) nbps_total: 0.);
}
else {
printf("# No sequences in the alignment have GR SS annotation.\n");
}
}
if(lfp != NULL) {
printf("# %d/%d sequences with %.3f individual bps on avg that conflict with SS_cons written to %s\n", nlist, msa->nseq, (float) nconflicts_list / (float) nlist, esl_opt_GetString(go, "-l"));
}
/* determine number of gaps per alignment column */
if((status = get_gaps_per_column(msa, &ngaps)) != eslOK) goto ERROR;
/* -x: determine max bp structure OR
* -a: list all conflicts in individual structures */
if(do_max || do_a) {
for(i = 1; i <= msa->alen; i++) {
if(nmates_l2r[i] > 1) {/* list the conflicts */
has_conflict[i] = TRUE;
for(j = 1; j <= msa->alen; j++) {
if(bp[i][j] > 0) {
if(do_a) printf("More than 1 right mates for left mate %4d %4d:%4d bp exists in %4d/%4d seqs (%.3f)\n", i, i, j, bp[i][j], msa->nseq - ngaps[i], (float) bp[i][j] / (float) (msa->nseq - ngaps[i]));
has_conflict[j] = TRUE;
}
}
}
}
for(i = 1; i <= msa->alen; i++) {
if(nmates_r2l[i] > 1) {/* list the conflicts */
has_conflict[i] = TRUE;
for(j = 1; j <= msa->alen; j++) {
if(bp[j][i] > 0) {
if(do_a) printf("More than 1 left mates for right mate %4d %4d:%4d bp exists in %4d/%4d seqs (%.3f)\n", i, j, i, bp[j][i], msa->nseq - ngaps[i], (float) bp[j][i] / (float) (msa->nseq - ngaps[i]));
has_conflict[j] = TRUE;
}
}
}
}
for(i = 1; i <= msa->alen; i++) {
/*printf("conflict[%4d]: %d\n", i, has_conflict[i]);*/
if(nmates_l2r[i] == 1 && (!(has_conflict[i]))) {
j = i+1;
while(bp[i][j] == 0) j++;
cons_ct[i] = j;
cons_ct[j] = i;
}
}
/* remove pseudoknotted bps greedily */
for(i = 1; i <= msa->alen; i++) {
j = cons_ct[i];
if(j != 0 && i < j) {
for(i2 = i+1; i2 <= msa->alen; i2++) {
j2 = cons_ct[i2];
if(j2 != 0 && i2 < j2) {
if((i2 < j) && (j < j2)) {
/*printf("KNOT %4d:%4d (%4d) %4d:%4d (%4d)\n", i, j, bp[i][j], i2, j2, bp[i2][j2]);*/
/* note: remove both if they have equal number of sequences */
if(bp[i][j] <= bp[i2][j2]) {
/*printf("rm %4d:%4d\n", i, j);*/
cons_ct[cons_ct[i]] = 0;
cons_ct[i] = 0;
}
if(bp[i][j] >= bp[i2][j2]) {
/*printf("rm %4d:%4d\n", i2, j2);*/
cons_ct[cons_ct[i2]] = 0;
cons_ct[i2] = 0;
}
}
}
}
}
}
}
/***************************************/
/*PARANOID, second check for knots
for(i = 1; i <= msa->alen; i++) {
j = cons_ct[i];
if(j != 0 && i < j) {
printf("BP: %4d:%4d\n", i, j);
for(i2 = 1; i2 <= msa->alen; i2++) {
j2 = cons_ct[i2];
if(j2 != 0 && i2 < j2) {
if((i2 < j) && (j < j2)) {
if((i < i2)) {
printf("KNOT %4d:%4d (%4d) %4d:%4d (%4d)\n", i, j, bp[i][j], i2, j2, bp[i2][j2]);
}
}
}
}
}
}
******************************************/
/***************************************/
/*PARANOID, check cons_ct for consistency
for(i = 1; i <= msa->alen; i++) {
if(cons_ct[i] != 0) {
if(cons_ct[cons_ct[i]] != i) { printf("ERROR: i: %4d cons_ct[i]: %4d cons_ct[cons_ct[i]]: %4d\n", i, cons_ct[i], cons_ct[cons_ct[i]]); }
}
}
*/
/*PARANOID, write out SS_cons
for(i = 1; i <= msa->alen; i++) {
if(i < cons_ct[i]) printf("<");
else if(cons_ct[i] != 0) { printf(">"); }
else printf(".");
}
printf("\n");
*/
/***************************************/
/* textize alignment */
if((status = esl_msa_Textize(msa)) != eslOK) esl_fatal("ERROR textizing alignment %d\n", nali);
/* --fmin */
if(do_fmin) {
/* define ss_cons */
prev_nbps = -1;
fthresh = 0.99;
inconsistent_flag = pknot_flag = FALSE;
printf("# Defining consensus structure:\n");
printf("# indi SS basepair aln columns i:j (from at least 1 indi SS) will become consensus basepair\n");
printf("# if > <x> individual SS contain i:j as a pair\n");
printf("# We'll search for minimal <x> that gives a consistent consensus structure.\n");
printf("# A consistent structure has each position involved in 0 or 1 basepairs.\n");
printf("#\n");
printf("# Alignment file: %s\n", alifile);
printf("# Alignment idx: %d\n", nali);
printf("# Number of seqs: %d\n", msa->nseq);
printf("#\n");
printf("# %5s %23s %6s\n", "<x>", "nseq-required-with-bp", "numbps");
printf("# %5s %23s %6s\n", "-----", "-----------------------", "------");
while(fthresh >= 0.00 && (inconsistent_flag == FALSE) && (pknot_flag == FALSE)) {
nbps = 0;
seqthresh = (int) (fthresh * msa->nseq);
/*printf("fthresh: %f seqthresh: %d nseq: %d\n", fthresh, seqthresh, msa->nseq);*/
esl_vec_ISet(cons_ct, msa->alen+1, 0);
for(i = 1; i <= msa->alen; i++) {
for(j = i+1; j <= msa->alen; j++) {
if(bp[i][j] > seqthresh) {
if(cons_ct[i] != 0 || cons_ct[j] != 0) {
inconsistent_flag = TRUE;
}
/* check for pseudoknots */
for(k = i+1; k < j; k++) {
l = cons_ct[k];
if((k < l) && (l > j)) {
pknot_flag = TRUE;
}
if((k > l) && (l != 0) && (l < i)) {
pknot_flag = TRUE;
}
}
cons_ct[i] = j;
cons_ct[j] = i;
nbps++;
}
}
}
if(inconsistent_flag)
printf(" %.3f %23d %s\n", fthresh, seqthresh+1, "inconsistent");
else if(pknot_flag)
printf(" %.3f %23d %s\n", fthresh, seqthresh+1, "pseudoknotted");
else {
if(nbps != prev_nbps) {
printf(" %.3f %23d %6d\n", fthresh, seqthresh+1, nbps);
}
fmin = fthresh;
}
fthresh -= 0.01;
prev_nbps = nbps;
}
fthresh = fmin;
esl_vec_ISet(cons_ct, msa->alen+1, 0);
}
/* --ffreq: determine structure by defining consensus bps that occur in <x> fraction of indi structures */
if(do_ffreq || do_fmin) {
if(do_fmin) { printf("#\n# <x> determined to be %.3f\n", fthresh); }
if(do_ffreq) {
printf("# Defining consensus structure:\n");
printf("# indi SS basepair aln columns i:j (from at least 1 indi SS) will become consensus basepair\n");
printf("# if > %f individual SS contain i:j as a pair\n", fthresh);
}
esl_vec_ISet(cons_ct, msa->alen+1, 0);
/* define ss_cons */
seqthresh = (int) (fthresh * msa->nseq);
/*printf("fthresh: %f seqthresh: %d nseq: %d\n", fthresh, seqthresh, msa->nseq);*/
for(i = 1; i <= msa->alen; i++) {
for(j = i+1; j <= msa->alen; j++) {
if(bp[i][j] > seqthresh) {
if(cons_ct[i] != 0) {
esl_fatal("ERROR, two base pairs including position %d satisfy threshold (%d:%d and %d:%d)!\n", i, i, cons_ct[i], i, j);
}
if(cons_ct[j] != 0) {
esl_fatal("ERROR, two base pairs including position %d satisfy threshold (%d:%d and %d:%d)!\n", j, j, cons_ct[j], i, j);
}
cons_ct[i] = j;
cons_ct[j] = i;
}
}
}
}
/* -r: redefine consensus struct by removing any bps that conflict with individual structures */
if(do_remove_bps) {
for(i = 1; i <= msa->alen; i++) {
if(!(removebp[i])) {
cons_ct[i] = xcons_ct[i];
cons_ct[cons_ct[i]] = i;
}
else {
printf("# Removing consensus bp: %d:%d\n", i, xcons_ct[i]);
cons_ct[xcons_ct[i]] = 0;
cons_ct[i] = 0;
}
}
}
/* -c: define consensus structure as indi sequence with highest number of consistent bps with structure OR */
/* --indi: define consensus structure as indi sequence <x> from --indi <x> */
if(do_consistent || do_indi2cons) {
if(do_indi2cons) {
indi = esl_opt_GetString(go, "--indi");
for(a = 0; a < msa->nseq; a++) {
if(strcmp(indi, msa->sqname[a]) == 0) break;
}
if(a == msa->nseq) esl_fatal("ERROR, could not find a sequence named %s in the alignment.\n", indi);
}
else { /* do_consistent */
a = max_nconsistent_aidx;
}
if(msa->ss == NULL || msa->ss[a] == NULL) esl_fatal("ERROR, no individual SS annotation for %s in the alignment.\n", msa->sqname[a]);
if((status = esl_wuss2ct(msa->ss[a], msa->alen, cons_ct)) != eslOK) {
esl_fatal("Second pass... SS annotation for sequence %d, aln %d is invalid.\n", (a), nali);
}
printf("# Defined new SS_cons as SS annotation for %s (%d basepairs)\n", msa->sqname[a], nbpsA[a]);
if(esl_opt_GetBoolean(go, "--rfc") || esl_opt_GetBoolean(go, "--rfindi")) {
if(msa->rf != NULL) { free(msa->rf); msa->rf = NULL; }
if((status = esl_strcat(&(msa->rf), -1, msa->aseq[a], msa->alen)) != eslOK) goto ERROR;
printf("# Defined new RF as %s sequence\n", msa->sqname[a]);
}
}
/* write out alignment with new SS_cons */
if(do_newcons) {
if((status = esl_ct2wuss(cons_ct, msa->alen, sscons)) != eslOK) goto ERROR;
if(msa->ss_cons != NULL) { free(msa->ss_cons); msa->ss_cons = NULL; }
if((status = esl_strcat(&(msa->ss_cons), -1, sscons, msa->alen)) != eslOK) goto ERROR;
status = esl_msafile_Write(ofp, msa, (esl_opt_GetBoolean(go, "--pfam") ? eslMSAFILE_PFAM : eslMSAFILE_STOCKHOLM));
if (status == eslEMEM) esl_fatal("Memory error when outputting alignment\n");
else if (status != eslOK) esl_fatal("Writing alignment file failed with error %d\n", status);
}
free(sscons);
free(cur_ct);
free(cons_ct);
free(xcons_ct);
for(i = 1; i <= msa->alen; i++) free(bp[i]);
free(bp);
esl_msa_Destroy(msa);
}
if (nali == 0) esl_fatal("No alignments found in file %s\n", alifile);
/* Cleanup, normal return
*/
if(lfp != NULL) fclose(lfp);
if(ofp != NULL) {
printf("# Alignment(s) saved to file %s\n", esl_opt_GetString(go, "-o"));
fclose(ofp);
}
esl_msafile_Close(afp);
esl_getopts_Destroy(go);
return 0;
ERROR:
if(afp) esl_msafile_Close(afp);
if(go) esl_getopts_Destroy(go);
if(msa) esl_msa_Destroy(msa);
if(lfp) fclose(lfp);
if(ofp) fclose(ofp);
esl_fatal("ERROR\n");
return 1;
}
int
main(int argc, char **argv)
{
ESL_ALPHABET *abc = NULL; /* sequence alphabet */
ESL_GETOPTS *go = NULL; /* command line processing */
ESL_RANDOMNESS *r = NULL; /* source of randomness */
P7_HMM *hmm = NULL; /* sampled HMM to emit from */
P7_HMM *core = NULL; /* safe copy of the HMM, before config */
P7_BG *bg = NULL; /* null model */
ESL_SQ *sq = NULL; /* sampled sequence */
P7_TRACE *tr = NULL; /* sampled trace */
P7_PROFILE *gm = NULL; /* profile */
int i,j;
int i1,i2;
int k1,k2;
int iseq;
FILE *fp = NULL;
double expected;
int do_ilocal;
char *hmmfile = NULL;
int nseq;
int do_swlike;
int do_ungapped;
int L;
int M;
int do_h2;
char *ipsfile = NULL;
char *kpsfile = NULL;
ESL_DMATRIX *imx = NULL;
ESL_DMATRIX *kmx = NULL;
ESL_DMATRIX *iref = NULL; /* reference matrix: expected i distribution under ideality */
int Lbins;
int status;
char errbuf[eslERRBUFSIZE];
/*****************************************************************
* Parse the command line
*****************************************************************/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal("Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal("Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
puts(usage);
puts("\n where options are:\n");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2 = indentation; 80=textwidth*/
return eslOK;
}
do_ilocal = esl_opt_GetBoolean(go, "-i");
hmmfile = esl_opt_GetString (go, "-m");
nseq = esl_opt_GetInteger(go, "-n");
do_swlike = esl_opt_GetBoolean(go, "-s");
do_ungapped = esl_opt_GetBoolean(go, "-u");
L = esl_opt_GetInteger(go, "-L");
M = esl_opt_GetInteger(go, "-M");
do_h2 = esl_opt_GetBoolean(go, "-2");
ipsfile = esl_opt_GetString (go, "--ips");
kpsfile = esl_opt_GetString (go, "--kps");
if (esl_opt_ArgNumber(go) != 0) {
puts("Incorrect number of command line arguments.");
printf("Usage: %s [options]\n", argv[0]);
return eslFAIL;
}
r = esl_randomness_CreateFast(0);
if (hmmfile != NULL)
{ /* Read the HMM (and get alphabet from it) */
P7_HMMFILE *hfp = NULL;
status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf);
if ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) != eslOK) {
if (status == eslEOD) esl_fatal("read failed, HMM file %s may be truncated?", hmmfile);
else if (status == eslEFORMAT) esl_fatal("bad file format in HMM file %s", hmmfile);
else if (status == eslEINCOMPAT) esl_fatal("HMM file %s contains different alphabets", hmmfile);
else esl_fatal("Unexpected error in reading HMMs");
}
M = hmm->M;
p7_hmmfile_Close(hfp);
}
else
{ /* Or sample the HMM (create alphabet first) */
abc = esl_alphabet_Create(eslAMINO);
if (do_ungapped) p7_hmm_SampleUngapped(r, M, abc, &hmm);
else if (do_swlike) p7_hmm_SampleUniform (r, M, abc, 0.05, 0.5, 0.05, 0.2, &hmm); /* tmi, tii, tmd, tdd */
else p7_hmm_Sample (r, M, abc, &hmm);
}
Lbins = M;
imx = esl_dmatrix_Create(Lbins, Lbins);
iref = esl_dmatrix_Create(Lbins, Lbins);
kmx = esl_dmatrix_Create(M, M);
esl_dmatrix_SetZero(imx);
esl_dmatrix_SetZero(iref);
esl_dmatrix_SetZero(kmx);
tr = p7_trace_Create();
sq = esl_sq_CreateDigital(abc);
bg = p7_bg_Create(abc);
core = p7_hmm_Clone(hmm);
if (do_h2) {
gm = p7_profile_Create(hmm->M, abc);
p7_H2_ProfileConfig(hmm, bg, gm, p7_UNILOCAL);
} else {
gm = p7_profile_Create(hmm->M, abc);
p7_ProfileConfig(hmm, bg, gm, L, p7_UNILOCAL);
if (p7_hmm_Validate (hmm, NULL, 0.0001) != eslOK) esl_fatal("whoops, HMM is bad!");
if (p7_profile_Validate(gm, NULL, 0.0001) != eslOK) esl_fatal("whoops, profile is bad!");
}
/* Sample endpoints.
* Also sample an ideal reference distribution for i endpoints. i
* endpoints are prone to discretization artifacts, when emitted
* sequences have varying lengths. Taking log odds w.r.t. an ideal
* reference that is subject to the same discretization artifacts
* cancels out the effect.
*/
for (iseq = 0; iseq < nseq; iseq++)
{
if (do_ilocal) ideal_local_endpoints (r, core, sq, tr, Lbins, &i1, &i2, &k1, &k2);
else profile_local_endpoints(r, core, gm, sq, tr, Lbins, &i1, &i2, &k1, &k2);
imx->mx[i1-1][i2-1] += 1.;
kmx->mx[k1-1][k2-1] += 1.;
/* reference distribution for i */
ideal_local_endpoints (r, core, sq, tr, Lbins, &i1, &i2, &k1, &k2);
iref->mx[i1-1][i2-1] += 1.;
}
/* Adjust both mx's to log_2(obs/exp) ratio */
printf("Before normalization/log-odds:\n");
printf(" i matrix values range from %f to %f\n", dmx_upper_min(imx), dmx_upper_max(imx));
printf(" k matrix values range from %f to %f\n", dmx_upper_min(kmx), dmx_upper_max(kmx));
printf("iref matrix values range from %f to %f\n", dmx_upper_min(iref), dmx_upper_max(iref));
expected = (double) nseq * 2. / (double) (M*(M+1));
for (i = 0; i < kmx->m; i++)
for (j = i; j < kmx->n; j++)
kmx->mx[i][j] = log(kmx->mx[i][j] / expected) / log(2.0);
for (i = 0; i < imx->m; i++)
for (j = i; j < imx->m; j++)
if (iref->mx[i][j] == 0. && imx->mx[i][j] == 0.)
imx->mx[i][j] = 0.;
else if (iref->mx[i][j] == 0.)
imx->mx[i][j] = eslINFINITY;
else if (imx->mx[i][j] == 0.)
imx->mx[i][j] = -eslINFINITY;
else
imx->mx[i][j] = log(imx->mx[i][j] / iref->mx[i][j]) / log(2.0);
/* Print ps files */
if (kpsfile != NULL) {
if ((fp = fopen(kpsfile, "w")) == NULL) esl_fatal("Failed to open output postscript file %s", kpsfile);
dmx_Visualize(fp, kmx, -4., 5.);
fclose(fp);
}
if (ipsfile != NULL) {
if ((fp = fopen(ipsfile, "w")) == NULL) esl_fatal("Failed to open output postscript file %s", ipsfile);
dmx_Visualize(fp, imx, -4., 5.);
/* dmx_Visualize(fp, imx, dmx_upper_min(imx), dmx_upper_max(imx)); */
fclose(fp);
}
printf("After normalization/log-odds:\n");
printf("i matrix values range from %f to %f\n", dmx_upper_min(imx), dmx_upper_max(imx));
printf("k matrix values range from %f to %f\n", dmx_upper_min(kmx), dmx_upper_max(kmx));
p7_profile_Destroy(gm);
p7_bg_Destroy(bg);
p7_hmm_Destroy(core);
p7_hmm_Destroy(hmm);
p7_trace_Destroy(tr);
esl_sq_Destroy(sq);
esl_dmatrix_Destroy(imx);
esl_dmatrix_Destroy(kmx);
esl_alphabet_Destroy(abc);
esl_randomness_Destroy(r);
esl_getopts_Destroy(go);
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line configuration */
struct cfg_s cfg; /* application configuration */
char *basename= NULL; /* base of the output file names */
char *alifile = NULL; /* alignment file name */
char *dbfile = NULL; /* name of seq db file */
char outfile[256]; /* name of an output file */
int alifmt; /* format code for alifile */
int dbfmt; /* format code for dbfile */
ESL_MSAFILE *afp = NULL; /* open alignment file */
ESL_MSA *origmsa = NULL; /* one multiple sequence alignment */
ESL_MSA *msa = NULL; /* MSA after frags are removed */
ESL_MSA *trainmsa= NULL; /* training set, aligned */
ESL_STACK *teststack=NULL; /* test set: stack of ESL_SQ ptrs */
int status; /* easel return code */
int nfrags; /* # of fragments removed */
int ntestdom; /* # of test domains */
int ntest; /* # of test sequences created */
int nali; /* number of alignments read */
double avgid;
/* Parse command line */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in app configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h")) cmdline_help(argv[0], go);
if (esl_opt_ArgNumber(go) != 3) cmdline_failure(argv[0], "Incorrect number of command line arguments\n");
basename = esl_opt_GetArg(go, 1);
alifile = esl_opt_GetArg(go, 2);
dbfile = esl_opt_GetArg(go, 3);
alifmt = eslMSAFILE_STOCKHOLM;
dbfmt = eslSQFILE_FASTA;
/* Set up the configuration structure shared amongst functions here */
if (esl_opt_IsDefault(go, "--seed")) cfg.r = esl_randomness_CreateTimeseeded();
else cfg.r = esl_randomness_Create(esl_opt_GetInteger(go, "--seed"));
cfg.abc = NULL; /* until we open the MSA file, below */
cfg.fragfrac = esl_opt_GetReal(go, "-F");
cfg.idthresh1 = esl_opt_GetReal(go, "-1");
cfg.idthresh2 = esl_opt_GetReal(go, "-2");
cfg.test_lens = NULL;
cfg.ntest = 0;
/* Open the output files */
if (snprintf(outfile, 256, "%s.msa", basename) >= 256) esl_fatal("Failed to construct output MSA file name");
if ((cfg.out_msafp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open MSA output file %s\n", outfile);
if (snprintf(outfile, 256, "%s.fa", basename) >= 256) esl_fatal("Failed to construct output FASTA file name");
if ((cfg.out_seqfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open FASTA output file %s\n", outfile);
if (snprintf(outfile, 256, "%s.pos", basename) >= 256) esl_fatal("Failed to construct pos test set summary file name");
if ((cfg.possummfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open pos test set summary file %s\n", outfile);
if (snprintf(outfile, 256, "%s.neg", basename) >= 256) esl_fatal("Failed to construct neg test set summary file name");
if ((cfg.negsummfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open neg test set summary file %s\n", outfile);
if (snprintf(outfile, 256, "%s.tbl", basename) >= 256) esl_fatal("Failed to construct benchmark table file name");
if ((cfg.tblfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open benchmark table file %s\n", outfile);
/* Open the MSA file; determine alphabet */
status = esl_msafile_Open(alifile, alifmt, NULL, &afp);
if (status == eslENOTFOUND) esl_fatal("Alignment file %s doesn't exist or is not readable\n", alifile);
else if (status == eslEFORMAT) esl_fatal("Couldn't determine format of alignment %s\n", alifile);
else if (status != eslOK) esl_fatal("Alignment file open failed with error %d\n", status);
if (esl_opt_GetBoolean(go, "--amino")) cfg.abc = esl_alphabet_Create(eslAMINO);
else if (esl_opt_GetBoolean(go, "--dna")) cfg.abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) cfg.abc = esl_alphabet_Create(eslRNA);
else {
int type;
status = esl_msafile_GuessAlphabet(afp, &type);
if (status == eslEAMBIGUOUS) esl_fatal("Failed to guess the bio alphabet used in %s.\nUse --dna, --rna, or --amino option to specify it.", alifile);
else if (status == eslEFORMAT) esl_fatal("Alignment file parse failed: %s\n", afp->errbuf);
else if (status == eslENODATA) esl_fatal("Alignment file %s is empty\n", alifile);
else if (status != eslOK) esl_fatal("Failed to read alignment file %s\n", alifile);
cfg.abc = esl_alphabet_Create(type);
}
esl_msafile_SetDigital(afp, cfg.abc);
if (cfg.abc->type == eslAMINO) esl_composition_SW34(cfg.fq);
else esl_vec_DSet(cfg.fq, cfg.abc->K, 1.0 / (double) cfg.abc->K);
/* Open and process the dbfile; make sure it's in the same alphabet */
process_dbfile(&cfg, dbfile, dbfmt);
/* Read and process MSAs one at a time */
nali = 0;
while ((status = esl_msa_Read(afp, &origmsa)) == eslOK)
{
remove_fragments(&cfg, origmsa, &msa, &nfrags);
separate_sets (&cfg, msa, &trainmsa, &teststack);
ntestdom = esl_stack_ObjectCount(teststack);
if (ntestdom >= 2)
{
esl_stack_Shuffle(cfg.r, teststack);
synthesize_positives(go, &cfg, msa->name, teststack, &ntest);
esl_msa_MinimGaps(trainmsa, NULL, NULL);
esl_msa_Write(cfg.out_msafp, trainmsa, eslMSAFILE_STOCKHOLM);
esl_dst_XAverageId(cfg.abc, trainmsa->ax, trainmsa->nseq, 10000, &avgid); /* 10000 is max_comparisons, before sampling kicks in */
fprintf(cfg.tblfp, "%-20s %3.0f%% %6d %6d %6d %6d %6d %6d\n", msa->name, 100.*avgid, (int) trainmsa->alen, msa->nseq, nfrags, trainmsa->nseq, ntestdom, ntest);
nali++;
}
esl_msa_Destroy(trainmsa);
esl_msa_Destroy(origmsa);
esl_msa_Destroy(msa);
}
if (status == eslEFORMAT) esl_fatal("Alignment file parse error, line %d of file %s:\n%s\nOffending line is:\n%s\n",
afp->linenumber, afp->fname, afp->errbuf, afp->buf);
else if (status != eslEOF) esl_fatal("Alignment file read failed with error code %d\n", status);
else if (nali == 0) esl_fatal("No alignments found in file %s\n", alifile);
if (nali > 0)
synthesize_negatives(go, &cfg, esl_opt_GetInteger(go, "-N"));
fclose(cfg.out_msafp);
fclose(cfg.out_seqfp);
fclose(cfg.possummfp);
fclose(cfg.negsummfp);
fclose(cfg.tblfp);
esl_randomness_Destroy(cfg.r);
esl_alphabet_Destroy(cfg.abc);
esl_msafile_Close(afp);
esl_getopts_Destroy(go);
return 0;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
ESL_ALPHABET *abc = NULL; /* biological alphabet */
char *alifile1= NULL; /* alignment 1 file name */
char *alifile2= NULL; /* alignment 2 file name */
int fmt; /* format code for alifiles */
ESLX_MSAFILE *afp1 = NULL; /* open alignment file 1 */
ESLX_MSAFILE *afp2 = NULL; /* open alignment file 2 */
ESL_MSA *msa1 = NULL; /* multiple sequence alignment 1 */
ESL_MSA *msa2 = NULL; /* multiple sequence alignment 2 */
int status; /* easel return code */
char errbuf[eslERRBUFSIZE*4];
int *msa1_to_msa2_map; /* map from <msafile1> to <msafile2> */
char *sub_msa1_to_msa2_mask; /* with --sub the map from <msafile1> to <msafile2> in mask form */
FILE *subfp = NULL;
/***********************************************
* Parse command line
***********************************************/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") )
{
esl_banner(stdout, argv[0], banner);
esl_usage (stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != 2)
{
printf("Incorrect number of command line arguments.\n");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
alifile1 = esl_opt_GetArg(go, 1);
alifile2 = esl_opt_GetArg(go, 2);
fmt = eslMSAFILE_STOCKHOLM;
/***********************************************
* Open the MSA files
***********************************************/
if (esl_opt_GetBoolean(go, "--amino")) abc = esl_alphabet_Create(eslAMINO);
else if (esl_opt_GetBoolean(go, "--dna")) abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) abc = esl_alphabet_Create(eslRNA);
if ( (status = eslx_msafile_Open(&abc, alifile1, NULL, fmt, NULL, &afp1)) != eslOK) eslx_msafile_OpenFailure(afp1, status);
if ( (status = eslx_msafile_Open(&abc, alifile2, NULL, fmt, NULL, &afp2)) != eslOK) eslx_msafile_OpenFailure(afp2, status);
/******************************************************************
* Read first alignment from each file, we only use the first one
******************************************************************/
if ((status = eslx_msafile_Read(afp1, &msa1)) != eslOK) eslx_msafile_ReadFailure(afp1, status);
if ((status = eslx_msafile_Read(afp2, &msa2)) != eslOK) eslx_msafile_ReadFailure(afp2, status);
/* map the alignments in msa1 and msa2 */
if(! esl_opt_IsOn(go, "--submap")) {
if((status = map_msas(go, errbuf, msa1, msa2, &msa1_to_msa2_map)) != eslOK) goto ERROR;
free(msa1_to_msa2_map);
}
/* --submap: if nec, map <msafile1> to a subset of it's own columns in <msafile2> */
else { /* --submap was enabled */
if ((subfp = fopen(esl_opt_GetString(go, "--submap"), "w")) == NULL)
ESL_FAIL(eslFAIL, errbuf, "Failed to open --submap output file %s\n", esl_opt_GetString(go, "--submap"));
if((status = map_sub_msas(go, errbuf, msa1, msa2, &sub_msa1_to_msa2_mask)) != eslOK) goto ERROR;
fprintf(subfp, "%s\n", sub_msa1_to_msa2_mask);
fclose(subfp);
subfp = NULL;
printf("# Mask of 1/0s with 1 indicating aln column in %s maps to a column in %s saved to file %s.\n", alifile1, alifile2, esl_opt_GetString(go, "--submap"));
free(sub_msa1_to_msa2_mask);
}
/* Cleanup, normal return
*/
eslx_msafile_Close(afp1);
eslx_msafile_Close(afp2);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
esl_msa_Destroy(msa1);
esl_msa_Destroy(msa2);
return 0;
ERROR:
if (afp1) eslx_msafile_Close(afp1);
if (afp2) eslx_msafile_Close(afp2);
if (go) esl_getopts_Destroy(go);
if (msa1) esl_msa_Destroy(msa1);
if (msa2) esl_msa_Destroy(msa2);
if (subfp) fclose(subfp);
esl_fatal(errbuf);
return 1; /* never reached */
}
static int
process_commandline(int argc, char **argv, ESL_GETOPTS **ret_go, char **ret_alifile, char **ret_postalifile)
{
ESL_GETOPTS *go = esl_getopts_Create(options);
int status;
if (esl_opt_ProcessEnvironment(go) != eslOK) { if (printf("Failed to process environment:\n%s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) { if (printf("Failed to parse command line:\n%s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_VerifyConfig(go) != eslOK) { if (printf("Failed to parse command line:\n%s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* help format: */
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
if (puts("\nBasic options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
if (puts("\nMask range options (format: --xxx 10-20,30-40 ) :") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
if (puts("\nOptions for selecting alphabet rather than guessing it:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
if (puts("\nAlternative model construction strategies:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 3, 2, 80);
if (puts("\nAlternative relative sequence weighting strategies:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
if (puts("\nOther options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 8, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) > 2) { if (puts("Incorrect number of command line arguments.") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_alifile = esl_opt_GetArg(go, 1)) == NULL) { if (puts("Failed to get <msafile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_IsUsed(go, "--alirange") || esl_opt_IsUsed(go, "--modelrange") ) {
if ((*ret_postalifile = esl_opt_GetArg(go, 2)) == NULL) { if (puts("Failed to get <postmsafile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
}
if (strcmp(*ret_alifile, "-") == 0 && ! esl_opt_IsOn(go, "--informat"))
{ if (puts("Must specify --informat to read <alifile> from stdin ('-')") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
*ret_go = go;
return eslOK;
FAILURE: /* all errors handled here are user errors, so be polite. */
esl_usage(stdout, argv[0], usage);
puts("\nwhere basic options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80);
printf("\nTo see more help on other available options, do:\n %s -h\n\n", argv[0]);
esl_getopts_Destroy(go);
exit(1);
ERROR:
if (go) esl_getopts_Destroy(go);
exit(status);
}
/**
* int main(int argc, char **argv)
* Main driver
*/
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line processing */
ESL_ALPHABET *abc = NULL;
char *hmmfile = NULL;
char *outhmmfile = NULL;
P7_HMMFILE *hfp = NULL;
FILE *outhmmfp; /* HMM output file handle */
P7_HMM *hmm = NULL;
P7_BG *bg = NULL;
int nhmm;
double x;
float KL;
int status;
char errbuf[eslERRBUFSIZE];
float average_internal_transitions[ p7H_NTRANSITIONS ];
int k;
char errmsg[eslERRBUFSIZE];
/* Process the command line options.
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
profillic_p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
puts("\nOptions:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=docgroup, 2 = indentation; 80=textwidth*/
exit(0);
}
if (esl_opt_ArgNumber(go) != 2)
{
puts("Incorrect number of command line arguments.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if ((hmmfile = esl_opt_GetArg(go, 1)) == NULL)
{
puts("Failed to read <input hmmfile> argument from command line.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if ((outhmmfile = esl_opt_GetArg(go, 2)) == NULL)
{
puts("Failed to read <output hmmfile> argument from command line.");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
profillic_p7_banner(stdout, argv[0], banner);
/* Initializations: open the input HMM file for reading
*/
status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf);
/* Initializations: open the output HMM file for writing
*/
if ((outhmmfp = fopen(outhmmfile, "w")) == NULL) ESL_FAIL(status, errmsg, "Failed to open HMM file %s for writing", outhmmfile);
/* Main body: read HMMs one at a time, print one line of stats
*/
printf("#\n");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s\n", "idx", "name", "accession", "nseq", "eff_nseq", "M", "relent", "info", "p relE", "compKL");
printf("# %-4s %-20s %-12s %8s %8s %6s %6s %6s %6s %6s\n", "----", "--------------------", "------------", "--------", "--------", "------", "------", "------", "------", "------");
nhmm = 0;
while ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) != eslEOF)
{
if (status == eslEOD) esl_fatal("read failed, HMM file %s may be truncated?", hmmfile);
else if (status == eslEFORMAT) esl_fatal("bad file format in HMM file %s", hmmfile);
else if (status == eslEINCOMPAT) esl_fatal("HMM file %s contains different alphabets", hmmfile);
else if (status != eslOK) esl_fatal("Unexpected error in reading HMMs from %s", hmmfile);
nhmm++;
if (bg == NULL) bg = p7_bg_Create(abc);
esl_vec_FSet(average_internal_transitions, p7H_NTRANSITIONS, 0.);
for( k = 1; k < hmm->M; k++ ) {
esl_vec_FAdd(average_internal_transitions, hmm->t[k], p7H_NTRANSITIONS);
}
// Match transitions
esl_vec_FNorm(average_internal_transitions, 3);
// Insert transitions
esl_vec_FNorm(average_internal_transitions + 3, 2);
// Delete transitions
esl_vec_FNorm(average_internal_transitions + 5, 2);
// Ok now set them.
for( k = 1; k < hmm->M; k++ ) {
esl_vec_FCopy( average_internal_transitions, p7H_NTRANSITIONS, hmm->t[k] );
}
if ((status = p7_hmm_Validate(hmm, errmsg, 0.0001)) != eslOK) return status;
if ((status = p7_hmmfile_WriteASCII(outhmmfp, -1, hmm)) != eslOK) ESL_FAIL(status, errmsg, "HMM save failed");
p7_MeanPositionRelativeEntropy(hmm, bg, &x);
p7_hmm_CompositionKLDist(hmm, bg, &KL, NULL);
printf("%-6d %-20s %-12s %8d %8.2f %6d %6.2f %6.2f %6.2f %6.2f\n",
nhmm,
hmm->name,
hmm->acc == NULL ? "-" : hmm->acc,
hmm->nseq,
hmm->eff_nseq,
hmm->M,
p7_MeanMatchRelativeEntropy(hmm, bg),
p7_MeanMatchInfo(hmm, bg),
x,
KL);
/* p7_MeanForwardScore(hmm, bg)); */
p7_hmm_Destroy(hmm);
}
p7_bg_Destroy(bg);
esl_alphabet_Destroy(abc);
p7_hmmfile_Close(hfp);
if (outhmmfp != NULL) fclose(outhmmfp);
esl_getopts_Destroy(go);
exit(0);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
ESL_STOPWATCH *w = esl_stopwatch_Create();
struct cfg_s cfg;
p7_Init();
/* Process command line options. */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
puts("\nCommon options:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1=docgroup, 2 = indentation; 80=textwidth*/
puts("\nChoice of score type :");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
puts("\nChoice of alignment mode :");
esl_opt_DisplayHelp(stdout, go, 3, 2, 80);
puts("\nChoice of multi vs. unihit config :");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
puts("\nChoice of generic vs. vector DP implementation :");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
puts("\nOutput options (use only in serial mode, for single HMM input):");
esl_opt_DisplayHelp(stdout, go, 6, 2, 80);
puts("\nControlling range of fitted tail masses :");
esl_opt_DisplayHelp(stdout, go, 7, 2, 80);
puts("\nRecalibrating E-values, and replacing HMM's existing parameters :");
esl_opt_DisplayHelp(stdout, go, 8, 2, 80);
puts("\nDebugging :");
esl_opt_DisplayHelp(stdout, go, 9, 2, 80);
puts("\nExperiments :");
esl_opt_DisplayHelp(stdout, go, 10, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != 1)
{
puts("Incorrect number of command line arguments.");
esl_usage(stdout, argv[0], usage);
puts("\nwhere general options are:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1=docgroup, 2 = indentation; 80=textwidth*/
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
/* Validate combinations of score/config/implementation (4x3x2x2, score x mode x hit x DPimpl: 20 of 48 possible combos valid */
if (esl_opt_GetBoolean(go, "--vector") && ! esl_opt_GetBoolean(go, "--local")) p7_Fail("SIMD vector implementations only work for local alignment."); /* -16/48 */
if (esl_opt_GetBoolean(go, "--msv") && ! esl_opt_GetBoolean(go, "--local")) p7_Fail("MSV scoring is local by definition: use --local."); /* -4/48 */
if (esl_opt_GetBoolean(go, "--vit") && ! esl_opt_GetBoolean(go, "--local")) p7_Fail("no p7_GViterbiDual for new dual-mode profile implemented yet"); /* -4/48 */
/* Initialize configuration shared across all kinds of masters
* and workers in this .c file.
*/
cfg.hmmfile = esl_opt_GetArg(go, 1);
cfg.r = esl_randomness_Create(esl_opt_GetInteger(go, "--seed"));
// cfg.abc = esl_alphabet_Create(eslAMINO);
cfg.my_rank = 0; /* MPI init will change this soon, if --mpi was set */
cfg.nproc = 0; /* MPI init will change this soon, if --mpi was set */
cfg.do_mpi = FALSE; /* --mpi will change this soon (below) if necessary */
cfg.do_stall = esl_opt_GetBoolean(go, "--stall");
cfg.N = esl_opt_GetInteger(go, "-N");
cfg.L = esl_opt_GetInteger(go, "-L");
cfg.hfp = NULL;
cfg.ofp = NULL;
cfg.survfp = NULL;
cfg.efp = NULL;
cfg.ffp = NULL;
cfg.xfp = NULL;
cfg.alfp = NULL;
cfg.bg = NULL;
/* This is our stall point, if we need to wait until we get a
* debugger attached to this process for debugging (especially
* useful for MPI):
*/
while (cfg.do_stall);
/* Start timing. */
esl_stopwatch_Start(w);
/* Main body:
* Handed off to serial version or MPI masters and workers as appropriate.
*/
#ifdef HAVE_MPI
if (esl_opt_GetBoolean(go, "--mpi"))
{
/* Initialize MPI, figure out who we are, and whether we're running
* this show (proc 0) or working in it (procs >0).
*/
cfg.do_mpi = TRUE;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &(cfg.my_rank));
MPI_Comm_size(MPI_COMM_WORLD, &(cfg.nproc));
if (cfg.my_rank == 0 && cfg.nproc < 2) p7_Fail("Need at least 2 MPI processes to run --mpi mode.");
if (cfg.my_rank > 0) mpi_worker(go, &cfg);
else mpi_master(go, &cfg);
esl_stopwatch_Stop(w);
esl_stopwatch_MPIReduce(w, 0, MPI_COMM_WORLD);
MPI_Finalize(); /* both workers and masters reach this line */
}
else
#endif /*HAVE_MPI*/
{
/* No MPI? Then we're just the serial master. */
serial_master(go, &cfg);
esl_stopwatch_Stop(w);
}
/* Stop timing. */
if (cfg.my_rank == 0) esl_stopwatch_Display(stdout, w, "# CPU time: ");
/* Clean up and exit. */
if (cfg.my_rank == 0) {
if (cfg.hfp != NULL) p7_hmmfile_Close(cfg.hfp);
if (esl_opt_IsOn(go, "-o")) fclose(cfg.ofp);
if (cfg.survfp != NULL) fclose(cfg.survfp);
if (cfg.efp != NULL) fclose(cfg.efp);
if (cfg.ffp != NULL) fclose(cfg.ffp);
if (cfg.xfp != NULL) fclose(cfg.xfp);
if (cfg.alfp != NULL) fclose(cfg.alfp);
}
p7_bg_Destroy(cfg.bg);
esl_alphabet_Destroy(cfg.abc);
esl_randomness_Destroy(cfg.r);
esl_getopts_Destroy(go);
esl_stopwatch_Destroy(w);
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go;
char *msafile;
ESLX_MSAFILE *afp;
ESL_MSA *msa;
int do_gsc;
int do_pb;
int do_blosum;
int maxN;
double maxid;
int nsmall, nbig;
int i;
int status;
/* Process command line */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal("%s", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal("%s", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE){
puts(usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
return 0;
}
do_blosum = esl_opt_GetBoolean(go, "--blosum");
do_gsc = esl_opt_GetBoolean(go, "--gsc");
do_pb = esl_opt_GetBoolean(go, "--pb");
maxid = esl_opt_GetReal (go, "--id");
maxN = esl_opt_GetInteger(go, "--maxN");
if (esl_opt_ArgNumber(go) != 1) {
puts("Incorrect number of command line arguments.");
puts(usage);
return 1;
}
if ((msafile = esl_opt_GetArg(go, 1)) == NULL) esl_fatal("%s", go->errbuf);
esl_getopts_Destroy(go);
/* Weight one or more alignments from input file
*/
if ((status = eslx_msafile_Open(NULL, msafile, NULL, eslMSAFILE_UNKNOWN, NULL, &afp)) != eslOK)
eslx_msafile_OpenFailure(afp, status);
while ( (status = eslx_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) eslx_msafile_ReadFailure(afp, status);
if (maxN > 0 && msa->nseq > maxN) { esl_msa_Destroy(msa); continue; }
if (do_gsc) esl_msaweight_GSC(msa);
else if (do_pb) esl_msaweight_PB(msa);
else if (do_blosum) esl_msaweight_BLOSUM(msa, maxid);
for (nsmall = 0, nbig = 0, i = 0; i < msa->nseq; i++) {
if (msa->wgt[i] < 0.2) nsmall++;
if (msa->wgt[i] > 5.0) nbig++;
}
printf("%-20s %5d %5d %8.4f %8.4f %5d %5d\n",
msa->name,
msa->nseq,
msa->alen,
esl_vec_DMin(msa->wgt, msa->nseq),
esl_vec_DMax(msa->wgt, msa->nseq),
nsmall,
nbig);
esl_msa_Destroy(msa);
}
eslx_msafile_Close(afp);
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_STOPWATCH *w;
ESL_GETOPTS *go;
char *msafile;
ESLX_MSAFILE *afp;
ESL_MSA *msa;
int do_gsc;
int do_pb;
int do_blosum;
int maxN;
double maxid;
double cpu;
int status;
/* Process command line
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal("failed to parse cmd line: %s", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal("failed to parse cmd line: %s", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") == TRUE) {
puts(usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, 0, 2, 80); /* 0=all docgroups; 2=indentation; 80=width */
return 0;
}
do_blosum = esl_opt_GetBoolean(go, "--blosum");
do_gsc = esl_opt_GetBoolean(go, "--gsc");
do_pb = esl_opt_GetBoolean(go, "--pb");
maxid = esl_opt_GetReal (go, "--id");
maxN = esl_opt_GetInteger(go, "--maxN");
if (esl_opt_ArgNumber(go) != 1) {
puts("Incorrect number of command line arguments.");
puts(usage);
return 1;
}
if ((msafile = esl_opt_GetArg(go, 1)) == NULL) esl_fatal("failed to parse cmd line: %s", go->errbuf);
esl_getopts_Destroy(go);
w = esl_stopwatch_Create();
/* Weight one or more alignments from input file
*/
if ((status = eslx_msafile_Open(NULL, msafile, NULL, eslMSAFILE_UNKNOWN, NULL, &afp)) != eslOK)
eslx_msafile_OpenFailure(afp, status);
while ( (status = eslx_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) eslx_msafile_ReadFailure(afp, status);
if (maxN > 0 && msa->nseq > maxN) { esl_msa_Destroy(msa); continue; }
esl_stopwatch_Start(w);
if (do_gsc) esl_msaweight_GSC(msa);
else if (do_pb) esl_msaweight_PB(msa);
else if (do_blosum) esl_msaweight_BLOSUM(msa, maxid);
esl_stopwatch_Stop(w);
cpu = w->user;
printf("%-20s %6d %6d %.3f\n", msa->name, msa->alen, msa->nseq, cpu);
esl_msa_Destroy(msa);
}
eslx_msafile_Close(afp);
esl_stopwatch_Destroy(w);
return eslOK;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
char *seqfile = NULL; /* sequence file name */
char *maskfile = NULL; /* mask coordinate file name */
int infmt = eslSQFILE_UNKNOWN; /* format code for seqfile */
int outfmt = eslSQFILE_FASTA; /* format code for output seqs */
ESL_SQFILE *sqfp = NULL; /* open sequence file */
ESL_FILEPARSER *maskefp = NULL; /* open mask coord file */
FILE *ofp = NULL; /* output stream for masked seqs */
char *source = NULL; /* name of current seq to mask */
char *p1, *p2; /* pointers used in parsing */
int64_t start, end; /* start, end coord for masking */
int64_t i, j, pos; /* coords in a sequence */
int64_t overmask; /* # of extra residues to mask */
ESL_SQ *sq = esl_sq_Create(); /* current sequence */
int do_fetching;
int do_lowercase;
int maskchar;
int status; /* easel return code */
/****************************************************************************
* Parse command line
****************************************************************************/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) cmdline_help (argv[0], go);
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
do_fetching = esl_opt_GetBoolean(go, "-R");
do_lowercase = esl_opt_GetBoolean(go, "-l");
overmask = (esl_opt_IsOn(go, "-x") ? esl_opt_GetInteger(go, "-x") : 0);
maskchar = (esl_opt_IsOn(go, "-m") ? esl_opt_GetChar(go, "-m") : 'X');
seqfile = esl_opt_GetArg(go, 1);
maskfile = esl_opt_GetArg(go, 2);
/* Open the <seqfile>: text mode, not digital */
if (esl_opt_GetString(go, "--informat") != NULL) {
infmt = esl_sqio_EncodeFormat(esl_opt_GetString(go, "--informat"));
if (infmt == eslSQFILE_UNKNOWN) cmdline_failure(argv[0], "%s is not a valid input sequence file format for --informat");
}
status = esl_sqfile_Open(seqfile, infmt, NULL, &sqfp);
if (status == eslENOTFOUND) cmdline_failure(argv[0], "Sequence file %s not found.\n", seqfile);
else if (status == eslEFORMAT) cmdline_failure(argv[0], "Format of file %s unrecognized.\n", seqfile);
else if (status == eslEINVAL) cmdline_failure(argv[0], "Can't autodetect stdin or .gz.\n");
else if (status != eslOK) cmdline_failure(argv[0], "Open failed, code %d.\n", status);
if(do_fetching)
{
status = esl_sqfile_OpenSSI(sqfp, NULL);
if (status == eslEFORMAT) cmdline_failure(argv[0], "SSI index is in incorrect format\n");
else if (status == eslERANGE) cmdline_failure(argv[0], "SSI index is in 64-bit format and we can't read it\n");
else if (status != eslOK) cmdline_failure(argv[0], "Failed to open SSI index\n");
}
/* Open the <maskfile> */
if (esl_fileparser_Open(maskfile, NULL, &maskefp) != eslOK)
cmdline_failure(argv[0], "Failed to open mask coordinate file %s\n", maskfile);
esl_fileparser_SetCommentChar(maskefp, '#');
/* Open the output file, if any */
if (esl_opt_GetString(go, "-o") != NULL)
{
if ((ofp = fopen(esl_opt_GetString(go, "-o"), "w")) == NULL)
cmdline_failure(argv[0], "Failed to open output file %s\n", esl_opt_GetString(go, "-o"));
}
else ofp = stdout;
/****************************************************************************
* Main loop over lines in <maskfile>
****************************************************************************/
/* Read one data line at a time from the <maskfile>;
* parse into data fields <seqname> <start> <end>
*/
while (esl_fileparser_NextLine(maskefp) == eslOK)
{
/* First field is sequence name */
if (esl_fileparser_GetTokenOnLine(maskefp, &source, NULL) != eslOK)
esl_fatal("Failed to read source seq name on line %d of file %s\n", maskefp->linenumber, maskfile);
/* Get the sequence */
if (do_fetching)
{ /* If the <seqfile> is SSI indexed, try to reposition it and read <source> seq by random access */
status = esl_sqio_Fetch(sqfp, source, sq);
if (status == eslENOTFOUND) esl_fatal("seq %s not found in SSI index for file %s\n", source, sqfp->filename);
else if (status == eslEINVAL) esl_fatal("No SSI index or can't reposition in file %s\n", sqfp->filename);
else if (status == eslEFORMAT) esl_fatal("Parse failed:\n%s\n", esl_sqfile_GetErrorBuf(sqfp));
else if (status != eslOK) esl_fatal("Unexpected failure in fetching %s from file %s\n", source, sqfp->filename);
}
else
{ /* else, assume we're reading sequentially; <sqfile> and <maskfile> have seqs in same order */
status = esl_sqio_Read(sqfp, sq);
if (status == eslEOF) esl_fatal("File %s ended prematurely; didn't find %s\n", sqfp->filename, source);
else if (status == eslEFORMAT) esl_fatal("Parse failed:\n%s\n", esl_sqfile_GetErrorBuf(sqfp));
else if (status != eslOK) esl_fatal("Unexpected error reading sequence file %s\n", sqfp->filename);
if ((strcmp(sq->name, source) != 0) && (strcmp(sq->acc, source) != 0))
esl_fatal("Sequences in <sqfile> and <maskfile> aren't in same order; try -R");
}
/* If we're masking by lowercase, first make sure everything's uppercase */
if (do_lowercase)
for (pos = 0; pos < sq->n; pos++)
if (isalpha(sq->seq[pos]))
sq->seq[pos] = toupper(sq->seq[pos]);
/* Next two fields are <start>, <end> for the masking */
/* possible future extension: wrap loop around this, enable multiple masked regions */
if (esl_fileparser_GetTokenOnLine(maskefp, &p1, NULL) != eslOK)
esl_fatal("Failed to read start coord on line %d of file %s\n", maskefp->linenumber, maskfile);
start = strtoll(p1, &p2, 0) - 1;
if (esl_fileparser_GetTokenOnLine(maskefp, &p2, NULL) != eslOK)
esl_fatal("Failed to read end coord on line %d of file %s\n", maskefp->linenumber, maskfile);
end = strtoll(p2, &p1, 0) - 1;
/* Do the masking */
if (esl_opt_GetBoolean(go, "-r")) /* Reverse masking */
{ /* leave start..end unmasked; mask prefix 0..start-1, end+1..L-1 */
i = 0;
j = ESL_MIN(sq->n-1, start - 1 + overmask);
for (pos = i; pos <= j; pos++)
if (isalpha(sq->seq[pos]))
sq->seq[pos] = (do_lowercase ? tolower(sq->seq[pos]) : maskchar);
i = ESL_MAX(0, end + 1 - overmask);
j = sq->n-1;
for (pos = i; pos <= j; pos++)
if (isalpha(sq->seq[pos]))
sq->seq[pos] = (do_lowercase ? tolower(sq->seq[pos]) : maskchar);
}
else
{ /* normal: mask start..end */
i = ESL_MAX(0, start - overmask);
j = ESL_MIN(sq->n-1, end + overmask);
for (pos = i; pos <= j; pos++)
if (isalpha(sq->seq[pos]))
sq->seq[pos] = (do_lowercase ? tolower(sq->seq[pos]) : maskchar);
}
esl_sqio_Write(ofp, sq, outfmt, FALSE);
esl_sq_Reuse(sq);
}
esl_sq_Destroy(sq);
esl_fileparser_Close(maskefp);
esl_sqfile_Close(sqfp);
esl_getopts_Destroy(go);
if (ofp != stdout) fclose(ofp);
return 0;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
char *seqfile = NULL;
ESL_SQFILE *sqfp = NULL;
int infmt = eslSQFILE_UNKNOWN;
int alphatype = eslUNKNOWN;
ESL_ALPHABET *abc = NULL;
ESL_SQ *sq = NULL;
int64_t nseq = 0;
int64_t nres = 0;
int64_t small = 0;
int64_t large = 0;
double *monoc = NULL; /* monoresidue composition per sequence */
double *monoc_all = NULL; /* monoresidue composition over all seqs */
int do_comp = FALSE;
int status = eslOK;
int wstatus;
int i;
int x;
/* Parse command line */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in app configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) cmdline_help(argv[0], go);
if (esl_opt_ArgNumber(go) != 1) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
seqfile = esl_opt_GetArg(go, 1);
do_comp = esl_opt_GetBoolean(go, "-c");
if (esl_opt_GetString(go, "--informat") != NULL) {
infmt = esl_sqio_FormatCode(esl_opt_GetString(go, "--informat"));
if (infmt == eslSQFILE_UNKNOWN) esl_fatal("%s is not a valid input sequence file format for --informat");
}
status = esl_sqfile_Open(seqfile, infmt, NULL, &sqfp);
if (status == eslENOTFOUND) esl_fatal("No such file %s", seqfile);
else if (status == eslEFORMAT) esl_fatal("Format of seqfile %s unrecognized.", seqfile);
else if (status != eslOK) esl_fatal("Open failed, code %d.", status);
if (esl_opt_GetBoolean(go, "--rna")) alphatype = eslRNA;
else if (esl_opt_GetBoolean(go, "--dna")) alphatype = eslDNA;
else if (esl_opt_GetBoolean(go, "--amino")) alphatype = eslAMINO;
else {
status = esl_sqfile_GuessAlphabet(sqfp, &alphatype);
if (status == eslEAMBIGUOUS) esl_fatal("Couldn't guess alphabet from first sequence in %s", seqfile);
else if (status == eslEFORMAT) esl_fatal("Sequence file parse error, line %d of file %s:\n%s\n",
sqfp->linenumber, seqfile, sqfp->errbuf);
else if (status == eslENODATA) esl_fatal("Sequence file %s contains no data?", seqfile);
else if (status != eslOK) esl_fatal("Failed to guess alphabet (error code %d)\n", status);
}
abc = esl_alphabet_Create(alphatype);
sq = esl_sq_CreateDigital(abc);
esl_sqfile_SetDigital(sqfp, abc);
if (do_comp) {
ESL_ALLOC(monoc, (abc->Kp) * sizeof(double));
ESL_ALLOC(monoc_all, (abc->Kp) * sizeof(double));
esl_vec_DSet(monoc_all, abc->Kp, 0.0);
esl_vec_DSet(monoc, abc->Kp, 0.0);
}
while ((wstatus = esl_sqio_ReadWindow(sqfp, 0, 4096, sq)) != eslEOF)
{
if (wstatus == eslOK)
{
if (do_comp)
for (i = 1; i <= sq->n; i++)
monoc[sq->dsq[i]]++;
}
else if (wstatus == eslEOD)
{
if (nseq == 0) { small = large = sq->L; }
else {
small = ESL_MIN(small, sq->L);
large = ESL_MAX(large, sq->L);
}
if (esl_opt_GetBoolean(go, "-a")) {
printf("= %-20s %8" PRId64 " %s\n", sq->name, sq->L, (sq->desc != NULL) ? sq->desc : "");
}
nres += sq->L;
nseq++;
esl_sq_Reuse(sq);
if (do_comp) {
esl_vec_DAdd(monoc_all, monoc, abc->Kp);
esl_vec_DSet(monoc, abc->Kp, 0.0);
}
}
else if (wstatus == eslEFORMAT)
{
esl_fatal("Failed to parse sequence at line %ld, file %s:\n%s",
(long) sqfp->linenumber, sqfp->filename, sqfp->errbuf);
}
else
esl_fatal("Failed in reading sequence:\n%s\n", sqfp->errbuf);
}
printf("Format: %s\n", esl_sqio_DescribeFormat(sqfp->format));
printf("Alphabet type: %s\n", esl_abc_DescribeType(abc->type));
printf("Number of sequences: %" PRId64 "\n", nseq);
printf("Total # residues: %" PRId64 "\n", nres);
printf("Smallest: %" PRId64 "\n", small);
printf("Largest: %" PRId64 "\n", large);
printf("Average length: %.1f\n", (float) nres / (float) nseq);
if (do_comp) {
printf("\nResidue composition:\n");
for (x = 0; x < abc->Kp; x++)
if (x < abc->K || monoc_all[x] > 0)
printf("residue: %c %10.0f %.4f\n", abc->sym[x], monoc_all[x], monoc_all[x] / (double) nres);
free(monoc);
free(monoc_all);
}
esl_alphabet_Destroy(abc);
esl_sq_Destroy(sq);
esl_sqfile_Close(sqfp);
esl_getopts_Destroy(go);
return 0;
ERROR:
return status;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
ESL_RANDOMNESS *r = NULL;
int nselect = 0;
char *filename = NULL;
FILE *fp = NULL;
char **larr = NULL;
char *buf = NULL;
int buflen = 0;
char *tmp = NULL;
int i,j;
int n;
/* Parse command line */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in app configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) cmdline_help(argv[0], go);
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
nselect = atoi(esl_opt_GetArg(go, 1));
filename = esl_opt_GetArg(go, 2);
r = esl_randomness_Create(esl_opt_GetInteger(go, "--seed"));
if ((larr = malloc(sizeof(char *) * nselect)) == NULL) esl_fatal("allocation failed");
if (strcmp(filename, "-") == 0) fp = stdin;
else {
if ((fp = fopen(filename, "r")) == NULL) esl_fatal("Failed to open file %s\n", filename);
}
n = 0;
while (esl_fgets(&buf, &buflen, fp) == eslOK)
{
n++;
i = esl_rnd_Roll(r, n);
if (i < nselect) {
for (j = i; j < nselect && j < n; j++)
{
tmp = larr[j];
larr[j] = buf;
buf = tmp;
}
free(buf);
buf = NULL;
buflen = 0;
}
}
for (i = 0; i < nselect; i++) printf("%s", larr[i]);
if (fp != stdin) fclose(fp);
for (i = 0; i < nselect; i++) free(larr[i]);
free(larr);
free(buf);
esl_randomness_Destroy(r);
esl_getopts_Destroy(go);
return 0;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line configuration */
struct cfg_s cfg; /* application configuration */
char *basename= NULL; /* base of the output file names */
char *alifile = NULL; /* alignment file name */
char *dbfile = NULL; /* name of seq db file */
char outfile[256]; /* name of an output file */
int alifmt; /* format code for alifile */
int dbfmt; /* format code for dbfile */
ESLX_MSAFILE *afp = NULL; /* open alignment file */
ESL_MSA *origmsa = NULL; /* one multiple sequence alignment */
ESL_MSA *msa = NULL; /* MSA after frags are removed */
ESL_MSA *trainmsa= NULL; /* training set, aligned */
ESL_STACK *teststack=NULL; /* test set: stack of ESL_SQ ptrs */
int status; /* easel return code */
int nfrags; /* # of fragments removed */
int ntestdom; /* # of test domains */
int ntest; /* # of test sequences created */
int nali; /* number of alignments read */
double avgid;
/* Parse command line */
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in app configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h")) cmdline_help(argv[0], go);
if (esl_opt_ArgNumber(go) != 3) cmdline_failure(argv[0], "Incorrect number of command line arguments\n");
basename = esl_opt_GetArg(go, 1);
alifile = esl_opt_GetArg(go, 2);
dbfile = esl_opt_GetArg(go, 3);
alifmt = eslMSAFILE_STOCKHOLM;
dbfmt = eslSQFILE_FASTA;
/* Set up the configuration structure shared amongst functions here */
if (esl_opt_IsDefault(go, "--seed")) cfg.r = esl_randomness_CreateTimeseeded();
else cfg.r = esl_randomness_Create(esl_opt_GetInteger(go, "--seed"));
cfg.abc = NULL; /* until we open the MSA file, below */
cfg.fragfrac = esl_opt_GetReal(go, "-F");
cfg.idthresh1 = esl_opt_GetReal(go, "-1");
cfg.idthresh2 = esl_opt_GetReal(go, "-2");
cfg.test_lens = NULL;
cfg.ntest = 0;
cfg.max_ntest = (esl_opt_IsOn(go, "--maxtest") ? esl_opt_GetInteger(go, "--maxtest") : 0);
cfg.max_ntrain = (esl_opt_IsOn(go, "--maxtrain") ? esl_opt_GetInteger(go, "--maxtrain") : 0);
/* Open the output files */
if (snprintf(outfile, 256, "%s.msa", basename) >= 256) esl_fatal("Failed to construct output MSA file name");
if ((cfg.out_msafp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open MSA output file %s\n", outfile);
if (snprintf(outfile, 256, "%s.fa", basename) >= 256) esl_fatal("Failed to construct output FASTA file name");
if ((cfg.out_seqfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open FASTA output file %s\n", outfile);
if (snprintf(outfile, 256, "%s.pos", basename) >= 256) esl_fatal("Failed to construct pos test set summary file name");
if ((cfg.possummfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open pos test set summary file %s\n", outfile);
if (snprintf(outfile, 256, "%s.neg", basename) >= 256) esl_fatal("Failed to construct neg test set summary file name");
if ((cfg.negsummfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open neg test set summary file %s\n", outfile);
if (snprintf(outfile, 256, "%s.tbl", basename) >= 256) esl_fatal("Failed to construct benchmark table file name");
if ((cfg.tblfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open benchmark table file %s\n", outfile);
if (esl_opt_GetBoolean(go, "--pid")) {
if (snprintf(outfile, 256, "%s.pid", basename) >= 256) esl_fatal("Failed to construct %%id table file name");
if ((cfg.pidfp = fopen(outfile, "w")) == NULL) esl_fatal("Failed to open %%id table file %s\n", outfile);
} else cfg.pidfp = NULL;
/* Open the MSA file, digital mode; determine alphabet */
if (esl_opt_GetBoolean(go, "--amino")) cfg.abc = esl_alphabet_Create(eslAMINO);
else if (esl_opt_GetBoolean(go, "--dna")) cfg.abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) cfg.abc = esl_alphabet_Create(eslRNA);
status = eslx_msafile_Open(&(cfg.abc), alifile, NULL, alifmt, NULL, &afp);
if (status != eslOK) eslx_msafile_OpenFailure(afp, status);
if (cfg.abc->type == eslAMINO) esl_composition_SW34(cfg.fq);
else esl_vec_DSet(cfg.fq, cfg.abc->K, 1.0 / (double) cfg.abc->K);
/* Open and process the dbfile; make sure it's in the same alphabet */
process_dbfile(&cfg, dbfile, dbfmt);
/* Read and process MSAs one at a time */
nali = 0;
while ((status = eslx_msafile_Read(afp, &origmsa)) != eslEOF)
{
if (status != eslOK) eslx_msafile_ReadFailure(afp, status);
esl_msa_ConvertDegen2X(origmsa);
esl_msa_Hash(origmsa);
remove_fragments(&cfg, origmsa, &msa, &nfrags);
separate_sets (&cfg, msa, &trainmsa, &teststack);
if ( esl_stack_ObjectCount(teststack) >= 2)
{
/* randomize test domain order, and apply size limit if any */
esl_stack_Shuffle(cfg.r, teststack);
if (cfg.max_ntest) pstack_select_topn(&teststack, cfg.max_ntest);
ntestdom = esl_stack_ObjectCount(teststack);
/* randomize training set alignment order, and apply size limit if any */
esl_msashuffle_PermuteSequenceOrder(cfg.r, trainmsa);
if (cfg.max_ntrain) msa_select_topn(&trainmsa, cfg.max_ntrain);
esl_msa_MinimGaps(trainmsa, NULL, NULL, FALSE);
if (esl_opt_GetBoolean(go, "--pid")) write_pids(cfg.pidfp, origmsa, trainmsa, teststack);
synthesize_positives(go, &cfg, msa->name, teststack, &ntest);
eslx_msafile_Write(cfg.out_msafp, trainmsa, eslMSAFILE_STOCKHOLM);
esl_dst_XAverageId(cfg.abc, trainmsa->ax, trainmsa->nseq, 10000, &avgid); /* 10000 is max_comparisons, before sampling kicks in */
fprintf(cfg.tblfp, "%-20s %3.0f%% %6d %6d %6d %6d %6d %6d\n", msa->name, 100.*avgid, (int) trainmsa->alen, msa->nseq, nfrags, trainmsa->nseq, ntestdom, ntest);
nali++;
}
esl_msa_Destroy(trainmsa);
esl_msa_Destroy(origmsa);
esl_msa_Destroy(msa);
}
if (nali == 0) esl_fatal("No alignments found in file %s\n", alifile);
synthesize_negatives(go, &cfg, esl_opt_GetInteger(go, "-N"));
fclose(cfg.out_msafp);
fclose(cfg.out_seqfp);
fclose(cfg.possummfp);
fclose(cfg.negsummfp);
fclose(cfg.tblfp);
if (cfg.pidfp) fclose(cfg.pidfp);
esl_randomness_Destroy(cfg.r);
esl_alphabet_Destroy(cfg.abc);
eslx_msafile_Close(afp);
esl_getopts_Destroy(go);
return 0;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
ESL_ALPHABET *nt_abc = esl_alphabet_Create(eslDNA);
ESL_ALPHABET *aa_abc = esl_alphabet_Create(eslAMINO);
ESL_GENCODE *gcode = NULL;
ESL_GENCODE_WORKSTATE *wrk = NULL;
char *seqfile = NULL;
int informat = eslSQFILE_UNKNOWN;
ESL_SQFILE *sqfp = NULL;
int status;
/*****************************************************************
* command line parsing
*****************************************************************/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK ||
esl_opt_VerifyConfig(go) != eslOK)
{
printf("Failed to parse command line: %s\n", go->errbuf);
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
if (esl_opt_GetBoolean(go, "-h") )
{
esl_banner(stdout, argv[0], banner);
esl_usage (stdout, argv[0], usage);
puts("\n where options are:");
esl_opt_DisplayHelp(stdout, go, /*docgroup=*/0, /*indent=*/2, /*textwidth=*/80);
puts("\nAvailable NCBI genetic code tables (for -c <id>):");
esl_gencode_DumpAltCodeTable(stdout);
exit(0);
}
if (esl_opt_ArgNumber(go) != 1)
{
printf("Incorrect number of command line arguments.\n");
esl_usage(stdout, argv[0], usage);
printf("\nTo see more help on available options, do %s -h\n\n", argv[0]);
exit(1);
}
seqfile = esl_opt_GetArg(go, 1);
if (esl_opt_IsOn(go, "--informat") &&
(informat = esl_sqio_EncodeFormat(esl_opt_GetString(go, "--informat"))) == eslMSAFILE_UNKNOWN)
esl_fatal("%s is not a valid input sequence file format for --informat", esl_opt_GetString(go, "--informat"));
status = esl_sqfile_OpenDigital(nt_abc, seqfile, informat, /*env=*/NULL, &sqfp);
if (status == eslENOTFOUND) esl_fatal("Failed to find (or open) sequence file %s", seqfile);
else if (status == eslEFORMAT) esl_fatal("Failed to recognize format of sequence file %s", seqfile);
else if (status != eslOK) esl_fatal("Failure in opening sequence file %s; code %d", seqfile, status);
/* A limitation. The esl_sqio_ReadWindow() interface needs to use SSI positioning
* to read reverse complement, and that doesn't work on nonrewindable streams.
*/
if ( esl_opt_GetBoolean(go, "-W") && ! esl_sqfile_IsRewindable(sqfp) && ! esl_opt_GetBoolean(go, "--watson"))
esl_fatal("esl-translate can't read reverse complement from a nonrewindable stream (stdin pipe, .gz file, etc).");
/* Set up the genetic code. Default = NCBI 1, the standard code; allow ORFs to start at any aa
*/
gcode = esl_gencode_Create(nt_abc, aa_abc);
esl_gencode_Set(gcode, esl_opt_GetInteger(go, "-c")); // default = 1, the standard genetic code
if (esl_opt_GetBoolean(go, "-m")) esl_gencode_SetInitiatorOnlyAUG(gcode);
else if (! esl_opt_GetBoolean(go, "-M")) esl_gencode_SetInitiatorAny(gcode); // note this is the default, if neither -m or -M are set
/* Set up the workstate structure, which contains both stateful
* info about our position in <sqfp> and the DNA <sq>, as well as
* one-time config info from options
*/
wrk = esl_gencode_WorkstateCreate(go, gcode);
/* The two styles of main processing loop:
*/
if (esl_opt_GetBoolean(go, "-W")) do_by_windows(gcode, wrk, sqfp);
else do_by_sequences(gcode, wrk, sqfp);
esl_gencode_WorkstateDestroy(wrk);
esl_sqfile_Close(sqfp);
esl_gencode_Destroy(gcode);
esl_alphabet_Destroy(aa_abc);
esl_alphabet_Destroy(nt_abc);
esl_getopts_Destroy(go);
return 0;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
char *alifile = NULL; /* alignment file name */
int infmt = eslMSAFILE_UNKNOWN; /* format code for alifile */
int outfmt = eslMSAFILE_UNKNOWN; /* output format for fetched msa's */
ESLX_MSAFILE *afp = NULL; /* open alignment file */
FILE *ofp = NULL; /* output stream for alignments */
int status; /* easel return code */
/***********************************************
* Parse command line
***********************************************/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) cmdline_help (argv[0], go);
if (esl_opt_ArgNumber(go) < 1) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
if (esl_opt_IsOn(go, "--informat")) {
infmt = eslx_msafile_EncodeFormat(esl_opt_GetString(go, "--informat"));
if (infmt == eslMSAFILE_UNKNOWN) esl_fatal("%s is not a valid input alignment file format for --informat", esl_opt_GetString(go, "--informat"));
}
outfmt = eslx_msafile_EncodeFormat(esl_opt_GetString(go, "--outformat"));
if (outfmt == eslMSAFILE_UNKNOWN) esl_fatal("%s is not a valid output alignment file format for --outformat", esl_opt_GetString(go, "--outformat"));
alifile = esl_opt_GetArg(go, 1);
/* Open the alignment file. */
if ( (status = eslx_msafile_Open(NULL, alifile, NULL, infmt, NULL, &afp)) != eslOK)
eslx_msafile_OpenFailure(afp, status);
/* Open the SSI index, if any */
if (! esl_opt_GetBoolean(go, "--index"))
{
if (afp->bf->mode_is == eslBUFFER_FILE ||
afp->bf->mode_is == eslBUFFER_ALLFILE ||
afp->bf->mode_is == eslBUFFER_MMAP)
{
char *ssifile = NULL;
esl_sprintf(&ssifile, "%s.ssi", afp->bf->filename);
status = esl_ssi_Open(ssifile, &(afp->ssi));
if (status == eslERANGE ) esl_fatal("SSI index %s has 64-bit offsets; this system doesn't support them", ssifile);
else if (status == eslEFORMAT) esl_fatal("SSI index %s has an unrecognized format. Try recreating, w/ esl-afetch --index", ssifile);
else if (status == eslENOTFOUND) afp->ssi = NULL;
else if (status != eslOK) esl_fatal("SSI index %s: open failed, error code %d\n", ssifile, status);
free(ssifile);
}
}
/* Open the output file, if any
*/
if (esl_opt_GetBoolean(go, "-O"))
{
if ((ofp = fopen(esl_opt_GetArg(go, 2), "w")) == NULL)
esl_fatal("Failed to open output file %s\n", esl_opt_GetArg(go, 2));
}
else if (esl_opt_GetString(go, "-o") != NULL)
{
if ((ofp = fopen(esl_opt_GetString(go, "-o"), "w")) == NULL)
esl_fatal("Failed to open output file %s\n", esl_opt_GetString(go, "-o"));
}
else ofp = stdout;
/* Hand off control flow as appropriate */
if (esl_opt_GetBoolean(go, "--index"))
{
if (esl_opt_ArgNumber(go) != 1) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
create_ssi_index(go, afp);
}
else if (esl_opt_GetBoolean(go, "-f"))
{
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
multifetch(go, ofp, outfmt, esl_opt_GetArg(go, 2), afp);
}
else
{
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
onefetch(go, ofp, outfmt, esl_opt_GetArg(go, 2), afp);
if (ofp != stdout) printf("\n\nRetrieved alignment %s.\n", esl_opt_GetArg(go, 2));
}
eslx_msafile_Close(afp);
esl_getopts_Destroy(go);
exit(0);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL;
int i, j;
int status = eslOK;
P7_HMMFILE *hfp = NULL; /* open input HMM file */
P7_HMM *hmm = NULL; /* one HMM query */
ESL_ALPHABET *abc = NULL; /* digital alphabet */
P7_BG *bg = NULL;
char errbuf[eslERRBUFSIZE];
char* hmmfile;
float *rel_ents = NULL;
float **heights = NULL;
float **probs = NULL;
float *ins_P = NULL;
float *ins_expL = NULL;
float *occupancy = NULL;
int mode = HMMLOGO_RELENT_ALL; //default
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) esl_fatal(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) esl_fatal(argv[0], "Error in configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) {
p7_banner (stdout, argv[0], banner);
esl_usage (stdout, argv[0], usage);
puts("\nOptions:");
esl_opt_DisplayHelp(stdout, go, 1, 2, 100);
exit(0);
}
if (esl_opt_ArgNumber(go) != 1) esl_fatal(argv[0], "Incorrect number of command line arguments.\n");
hmmfile = esl_opt_GetArg(go, 1);
if (esl_opt_IsOn(go, "--height_relent_all"))
mode = HMMLOGO_RELENT_ALL;
else if (esl_opt_IsOn(go, "--height_relent_abovebg"))
mode = HMMLOGO_RELENT_ABOVEBG;
else if (esl_opt_IsOn(go, "--height_score"))
mode = HMMLOGO_SCORE;
else
mode = HMMLOGO_RELENT_ALL; //default
/* Open the query profile HMM file */
status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf);
status = p7_hmmfile_Read(hfp, &abc, &hmm);
bg = p7_bg_Create(abc);
ESL_ALLOC(rel_ents, (hmm->M+1) * sizeof(float));
ESL_ALLOC(heights, (hmm->M+1) * sizeof(float*));
ESL_ALLOC(probs, (hmm->M+1) * sizeof(float*));
for (i = 1; i <= hmm->M; i++) {
ESL_ALLOC(heights[i], abc->K * sizeof(float));
ESL_ALLOC(probs[i], abc->K * sizeof(float));
}
/* residue heights */
if (mode == HMMLOGO_RELENT_ALL) {
printf ("max expected height = %.2f\n", hmmlogo_maxHeight(bg) );
hmmlogo_RelativeEntropy_all(hmm, bg, rel_ents, probs, heights);
} else if (mode == HMMLOGO_RELENT_ABOVEBG) {
printf ("max expected height = %.2f\n", hmmlogo_maxHeight(bg) );
hmmlogo_RelativeEntropy_above_bg(hmm, bg, rel_ents, probs, heights);
} else if (mode == HMMLOGO_SCORE) {
hmmlogo_ScoreHeights(hmm, bg, heights );
}
printf ("Residue heights\n");
for (i = 1; i <= hmm->M; i++) {
printf("%d: ", i);
for (j=0; j<abc->K; j++)
printf("%6.3f ", heights[i][j] );
if (mode != HMMLOGO_SCORE)
printf(" (%6.3f)", rel_ents[i]);
printf("\n");
}
if (rel_ents != NULL) free(rel_ents);
if (heights != NULL) {
for (i = 1; i <= hmm->M; i++)
if (heights[i] != NULL) free(heights[i]);
free(heights);
}
/* indel values */
if (! esl_opt_IsOn(go, "--no_indel")) {
ESL_ALLOC(ins_P, (hmm->M+1) * sizeof(float));
ESL_ALLOC(ins_expL, (hmm->M+1) * sizeof(float));
ESL_ALLOC(occupancy, (hmm->M+1) * sizeof(float));
hmmlogo_IndelValues(hmm, ins_P, ins_expL, occupancy);
printf ("Indel values\n");
for (i = 1; i <= hmm->M; i++)
printf("%d: %6.3f %6.3f %6.3f\n", i, ins_P[i], ins_expL[i], occupancy[i] );
free(ins_P);
free(ins_expL);
free(occupancy);
}
p7_hmmfile_Close(hfp);
esl_alphabet_Destroy(abc);
p7_bg_Destroy(bg);
exit(0);
ERROR:
if (rel_ents != NULL) free(rel_ents);
if (heights != NULL) {
for (i = 1; i <= hmm->M; i++)
if (heights[i] != NULL) free(heights[i]);
free(heights);
}
if (hfp != NULL) p7_hmmfile_Close(hfp);
if (abc != NULL) esl_alphabet_Destroy(abc);
if (ins_P != NULL) free(ins_P);
if (ins_expL != NULL) free(ins_expL);
if (occupancy != NULL) free(occupancy);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* command line processing */
ESL_ALPHABET *abc = NULL; /* sequence alphabet */
ESL_RANDOMNESS *r = NULL; /* source of randomness */
char *hmmfile = NULL; /* file to read HMM(s) from */
P7_HMMFILE *hfp = NULL; /* open hmmfile */
P7_HMM *hmm = NULL; /* HMM to emit from */
FILE *ofp = NULL; /* output stream */
int outfmt = 0;
int nhmms = 0;
int status;
char errbuf[eslERRBUFSIZE];
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h")) cmdline_help (argv[0], go);
if (esl_opt_ArgNumber(go) != 1) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
if ((hmmfile = esl_opt_GetArg(go, 1)) == NULL) cmdline_failure(argv[0], "Failed to get <hmmfile> on cmdline: %s\n", go->errbuf);
if ( esl_opt_IsOn(go, "-o") ) {
if ((ofp = fopen(esl_opt_GetString(go, "-o"), "w")) == NULL) esl_fatal("Failed to open output file %s", esl_opt_GetString(go, "-o"));
} else ofp = stdout;
if (esl_opt_GetBoolean(go, "-a")) outfmt = eslMSAFILE_STOCKHOLM;
else outfmt = eslSQFILE_FASTA;
r = esl_randomness_CreateFast(esl_opt_GetInteger(go, "--seed"));
status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf);
while ((status = p7_hmmfile_Read(hfp, &abc, &hmm)) != eslEOF)
{
if (status == eslEFORMAT) esl_fatal("Bad file format in HMM file %s:\n%s\n", hfp->fname, hfp->errbuf);
else if (status == eslEINCOMPAT) esl_fatal("HMM in %s is not in the expected %s alphabet\n", hfp->fname, esl_abc_DecodeType(abc->type));
else if (status != eslOK) esl_fatal("Unexpected error in reading HMMs from %s\n", hfp->fname);
nhmms++;
if (esl_opt_GetBoolean(go, "-c")) emit_consensus(go, ofp, outfmt, hmm);
else if (esl_opt_GetBoolean(go, "-C")) emit_fancycons(go, ofp, outfmt, hmm);
else if (esl_opt_GetBoolean(go, "-a")) emit_alignment(go, ofp, outfmt, r, hmm);
else emit_sequences(go, ofp, outfmt, r, hmm);
p7_hmm_Destroy(hmm);
}
if (nhmms == 0) esl_fatal("Empty HMM file %s? No HMM data found.\n");
if (esl_opt_IsOn(go, "-o")) { fclose(ofp); }
esl_randomness_Destroy(r);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
p7_hmmfile_Close(hfp);
return eslOK;
}
/* process_commandline()
*
* Processes the commandline, filling in fields in <cfg> and creating and returning
* an <ESL_GETOPTS> options structure. The help page (hmmsearch -h) is formatted
* here.
*/
static int
process_commandline(int argc, char **argv, ESL_GETOPTS **ret_go, char **ret_hmmfile, char **ret_seqfile)
{
ESL_GETOPTS *go = esl_getopts_Create(options);
int status;
if (esl_opt_ProcessEnvironment(go) != eslOK) { if (printf("Failed to process environment: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) { if (printf("Failed to parse command line: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if (esl_opt_VerifyConfig(go) != eslOK) { if (printf("Failed to parse command line: %s\n", go->errbuf) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* help format: */
if (esl_opt_GetBoolean(go, "-h") == TRUE)
{
p7_banner(stdout, argv[0], banner);
esl_usage(stdout, argv[0], usage);
if (puts("\nBasic options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1= group; 2 = indentation; 80=textwidth*/
if (puts("\nOptions controlling output:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 2, 2, 80);
if (puts("\nOptions controlling reporting thresholds:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 4, 2, 80);
if (puts("\nOptions controlling inclusion (significance) thresholds:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 5, 2, 80);
if (puts("\nOptions for model-specific thresholding:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 6, 2, 80);
if (puts("\nOptions controlling acceleration heuristics:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 7, 2, 80);
if (puts("\nOther expert options:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 12, 2, 80);
exit(0);
}
if (esl_opt_ArgNumber(go) != 2) { if (puts("Incorrect number of command line arguments.") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_hmmfile = esl_opt_GetArg(go, 1)) == NULL) { if (puts("Failed to get <hmmdb> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
if ((*ret_seqfile = esl_opt_GetArg(go, 2)) == NULL) { if (puts("Failed to get <seqfile> argument on command line") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
/* Validate any attempted use of stdin streams */
if (strcmp(*ret_hmmfile, "-") == 0)
{ if (puts("nhmmscan cannot read <hmm database> from stdin stream, because it must have hmmpress'ed auxfiles") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto FAILURE; }
*ret_go = go;
return eslOK;
FAILURE: /* all errors handled here are user errors, so be polite. */
esl_usage(stdout, argv[0], usage);
if (puts("\nwhere most common options are:") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_opt_DisplayHelp(stdout, go, 1, 2, 80); /* 1= group; 2 = indentation; 80=textwidth*/
if (printf("\nTo see more help on available options, do %s -h\n\n", argv[0]) < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
esl_getopts_Destroy(go);
exit(1);
ERROR:
if (go) esl_getopts_Destroy(go);
exit(status);
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = NULL; /* application configuration */
char *hmmfile = NULL; /* HMM file name */
char *seqfile = NULL; /* sequence file name */
char *mapfile = NULL; /* optional mapped MSA file name */
int infmt = eslSQFILE_UNKNOWN;
int outfmt = eslMSAFILE_STOCKHOLM;
P7_HMMFILE *hfp = NULL; /* open HMM file */
ESL_SQFILE *sqfp = NULL; /* open sequence file */
char *outfile = NULL; /* output filename */
FILE *ofp = stdout; /* output stream */
ESL_SQ **sq = NULL; /* array of sequences */
void *p = NULL; /* tmp ptr for reallocation */
int nseq = 0; /* # of sequences in <seqfile> */
int mapseq = 0; /* # of sequences in mapped MSA */
int totseq = 0; /* # of seqs in all sources */
ESL_ALPHABET *abc = NULL; /* alphabet (set from the HMM file)*/
P7_HMM *hmm = NULL;
P7_TRACE **tr = NULL; /* array of tracebacks */
ESL_MSA *msa = NULL; /* resulting multiple alignment */
int msaopts = 0; /* flags to p7_tracealign_Seqs() */
int idx; /* counter over seqs, traces */
int status; /* easel/hmmer return code */
char errbuf[eslERRBUFSIZE];
/* Parse the command line
*/
go = esl_getopts_Create(options);
if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf);
if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in configuration: %s\n", go->errbuf);
if (esl_opt_GetBoolean(go, "-h") ) cmdline_help (argv[0], go);
if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n");
hmmfile = esl_opt_GetArg(go, 1);
seqfile = esl_opt_GetArg(go, 2);
if (strcmp(hmmfile, "-") == 0 && strcmp(seqfile, "-") == 0)
cmdline_failure(argv[0], "Either <hmmfile> or <seqfile> may be '-' (to read from stdin), but not both.\n");
msaopts |= p7_ALL_CONSENSUS_COLS; /* default as of 3.1 */
if (esl_opt_GetBoolean(go, "--trim")) msaopts |= p7_TRIM;
/* If caller declared an input format, decode it
*/
if (esl_opt_IsOn(go, "--informat")) {
infmt = esl_sqio_EncodeFormat(esl_opt_GetString(go, "--informat"));
if (infmt == eslSQFILE_UNKNOWN) cmdline_failure(argv[0], "%s is not a recognized input sequence file format\n", esl_opt_GetString(go, "--informat"));
}
/* Determine output alignment file format */
outfmt = eslx_msafile_EncodeFormat(esl_opt_GetString(go, "--outformat"));
if (outfmt == eslMSAFILE_UNKNOWN) cmdline_failure(argv[0], "%s is not a recognized output MSA file format\n", esl_opt_GetString(go, "--outformat"));
/* Open output stream */
if ( (outfile = esl_opt_GetString(go, "-o")) != NULL)
{
if ((ofp = fopen(outfile, "w")) == NULL)
cmdline_failure(argv[0], "failed to open -o output file %s for writing\n", outfile);
}
/* If caller forced an alphabet on us, create the one the caller wants
*/
if (esl_opt_GetBoolean(go, "--amino")) abc = esl_alphabet_Create(eslAMINO);
else if (esl_opt_GetBoolean(go, "--dna")) abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--rna")) abc = esl_alphabet_Create(eslRNA);
/* Read one HMM, and make sure there's only one.
*/
status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf);
if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf);
else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf);
status = p7_hmmfile_Read(hfp, &abc, &hmm);
if (status == eslEFORMAT) p7_Fail("Bad file format in HMM file %s:\n%s\n", hfp->fname, hfp->errbuf);
else if (status == eslEINCOMPAT) p7_Fail("HMM in %s is not in the expected %s alphabet\n", hfp->fname, esl_abc_DecodeType(abc->type));
else if (status == eslEOF) p7_Fail("Empty HMM file %s? No HMM data found.\n", hfp->fname);
else if (status != eslOK) p7_Fail("Unexpected error in reading HMMs from %s\n", hfp->fname);
status = p7_hmmfile_Read(hfp, &abc, NULL);
if (status != eslEOF) p7_Fail("HMM file %s does not contain just one HMM\n", hfp->fname);
p7_hmmfile_Close(hfp);
/* We're going to build up two arrays: sequences and traces.
* If --mapali option is chosen, the first set of sequences/traces is from the provided alignment
*/
if ( (mapfile = esl_opt_GetString(go, "--mapali")) != NULL)
{
map_alignment(mapfile, hmm, &sq, &tr, &mapseq);
}
totseq = mapseq;
/* Read digital sequences into an array (possibly concat'ed onto mapped seqs)
*/
status = esl_sqfile_OpenDigital(abc, seqfile, infmt, NULL, &sqfp);
if (status == eslENOTFOUND) p7_Fail("Failed to open sequence file %s for reading\n", seqfile);
else if (status == eslEFORMAT) p7_Fail("Sequence file %s is empty or misformatted\n", seqfile);
else if (status != eslOK) p7_Fail("Unexpected error %d opening sequence file %s\n", status, seqfile);
ESL_RALLOC(sq, p, sizeof(ESL_SQ *) * (totseq + 1));
sq[totseq] = esl_sq_CreateDigital(abc);
nseq = 0;
while ((status = esl_sqio_Read(sqfp, sq[totseq+nseq])) == eslOK)
{
nseq++;
ESL_RALLOC(sq, p, sizeof(ESL_SQ *) * (totseq+nseq+1));
sq[totseq+nseq] = esl_sq_CreateDigital(abc);
}
if (status == eslEFORMAT) esl_fatal("Parse failed (sequence file %s):\n%s\n",
sqfp->filename, esl_sqfile_GetErrorBuf(sqfp));
else if (status != eslEOF) esl_fatal("Unexpected error %d reading sequence file %s", status, sqfp->filename);
esl_sqfile_Close(sqfp);
totseq += nseq;
/* Remaining initializations, including trace array allocation
*/
ESL_RALLOC(tr, p, sizeof(P7_TRACE *) * totseq);
for (idx = mapseq; idx < totseq; idx++)
tr[idx] = p7_trace_CreateWithPP();
p7_tracealign_computeTraces(hmm, sq, mapseq, totseq - mapseq, tr);
p7_tracealign_Seqs(sq, tr, totseq, hmm->M, msaopts, hmm, &msa);
eslx_msafile_Write(ofp, msa, outfmt);
for (idx = 0; idx <= totseq; idx++) esl_sq_Destroy(sq[idx]); /* including sq[nseq] because we overallocated */
for (idx = 0; idx < totseq; idx++) p7_trace_Destroy(tr[idx]);
free(sq);
free(tr);
esl_msa_Destroy(msa);
p7_hmm_Destroy(hmm);
if (ofp != stdout) fclose(ofp);
esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
return eslOK;
ERROR:
return status;
}