/* onefetch():
* Given one <key> (an MSA name or accession), retrieve the corresponding MSA.
* In SSI mode, we can do this quickly by positioning the file, then regurgitating
* every line until the end-of-alignment marker; we don't even have to parse.
* Without an SSI index, we have to parse the MSAs sequentially 'til we find
* the one we're after.
*/
static void
onefetch(ESL_GETOPTS *go, FILE *ofp, int outfmt, char *key, ESLX_MSAFILE *afp)
{
ESL_MSA *msa = NULL;
int nali = 1;
int status;
if (afp->ssi)
{
status = eslx_msafile_PositionByKey(afp, key);
if (status == eslENOTFOUND) esl_fatal("MSA %s not found in SSI index for file %s\n", key, afp->bf->filename);
else if (status == eslEFORMAT) esl_fatal("Failed to parse SSI index for %s\n", afp->bf->filename);
else if (status != eslOK) esl_fatal("Failed to look up location of MSA %s in SSI index of file %s\n", key, afp->bf->filename);
if ( (afp->format == eslMSAFILE_STOCKHOLM && outfmt == eslMSAFILE_STOCKHOLM) ||
(afp->format == eslMSAFILE_PFAM && outfmt == eslMSAFILE_PFAM))
{
regurgitate_one_stockholm_entry(ofp, afp);
}
else
{
if ((status = eslx_msafile_Read(afp, &msa)) != eslOK)
eslx_msafile_ReadFailure(afp, status);
eslx_msafile_Write(ofp, msa, outfmt);
esl_msa_Destroy(msa);
}
}
else
{ /* without an index, we have to brute-force search the file */
while ((status = eslx_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) eslx_msafile_ReadFailure(afp, status);
if (! msa->name)
esl_fatal("Every alignment in file must have a name to be retrievable. Failed to find name of alignment #%d\n", nali);
if (strcmp(key, msa->name) == 0 || (msa->acc != NULL && strcmp(key, msa->acc) == 0))
break;
nali++;
esl_msa_Destroy(msa);
}
if (! msa) esl_fatal("Failed to find alignment %s\n", key);
eslx_msafile_Write(ofp, msa, outfmt);
esl_msa_Destroy(msa);
}
}
static void
emit_alignment(ESL_GETOPTS *go, FILE *ofp, int outfmt, ESL_RANDOMNESS *r, P7_HMM *hmm)
{
ESL_MSA *msa = NULL;
ESL_SQ **sq = NULL;
P7_TRACE **tr = NULL;
int N = esl_opt_GetInteger(go, "-N");
int optflags = p7_ALL_CONSENSUS_COLS;
int i;
if ((tr = malloc(sizeof(P7_TRACE *) * N)) == NULL) esl_fatal("failed to allocate trace array");
if ((sq = malloc(sizeof(ESL_SQ *) * N)) == NULL) esl_fatal("failed to allocate seq array");
for (i = 0; i < N; i++)
{
if ((sq[i] = esl_sq_CreateDigital(hmm->abc)) == NULL) esl_fatal("failed to allocate seq");
if ((tr[i] = p7_trace_Create()) == NULL) esl_fatal("failed to allocate trace");
}
for (i = 0; i < N; i++)
{
if (p7_CoreEmit(r, hmm, sq[i], tr[i]) != eslOK) esl_fatal("Failed to emit sequence");
if (esl_sq_FormatName(sq[i], "%s-sample%d", hmm->name, i+1) != eslOK) esl_fatal("Failed to set sequence name\n");
}
p7_tracealign_Seqs(sq, tr, N, hmm->M, optflags, hmm, &msa);
eslx_msafile_Write(ofp, msa, outfmt);
for (i = 0; i < N; i++) p7_trace_Destroy(tr[i]); free(tr);
for (i = 0; i < N; i++) esl_sq_Destroy(sq[i]); free(sq);
esl_msa_Destroy(msa);
return;
}
int
main(int argc, char **argv)
{
ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 1, argc, argv, banner, usage);
ESL_RANDOMNESS *rng = esl_randomness_Create(0);
char *msafile = esl_opt_GetArg(go, 1);
int fmt = eslMSAFILE_UNKNOWN;
ESL_ALPHABET *abc = NULL;
ESLX_MSAFILE *afp = NULL;
ESL_MSA *msa = NULL;
int textmode = esl_opt_GetBoolean(go, "--text");
int nali = 0;
int status;
/* If you know the alphabet you want, create it - you'll pass it to eslx_msafile_Open() */
if (esl_opt_GetBoolean(go, "--rna")) abc = esl_alphabet_Create(eslRNA);
else if (esl_opt_GetBoolean(go, "--dna")) abc = esl_alphabet_Create(eslDNA);
else if (esl_opt_GetBoolean(go, "--amino")) abc = esl_alphabet_Create(eslAMINO);
/* Open in text or digital mode.
* To let the Open() function autoguess the format, you pass <infmt=eslMSAFILE_UNKNOWN>.
* To let it autoguess the alphabet, you set <abc=NULL> and pass <&abc>.
* To open in text mode instead of digital, you pass <NULL> for the alphabet argument.
* eslx_msafile_OpenFailure() is a convenience, printing various diagnostics of any
* open failure to <stderr>. You can of course handle your own diagnostics instead.
*/
if (textmode) status = eslx_msafile_Open(NULL, msafile, NULL, fmt, NULL, &afp);
else status = eslx_msafile_Open(&abc, msafile, NULL, fmt, NULL, &afp);
if (status != eslOK) eslx_msafile_OpenFailure(afp, status);
fmt = afp->format;
while ((status = eslx_msafile_Read(afp, &msa)) == eslOK)
{
/* if digital MSA: msa->ax[idx=0..nseq-1][acol=1..alen] is the alignment data;
* if text MSA: msa->aseq[idx=0..nseq-1][acol=0..alen-1] */
nali++;
/* permute it */
esl_msashuffle_PermuteSequenceOrder(rng, msa);
eslx_msafile_Write(stdout, msa, fmt);
esl_msa_Destroy(msa);
}
if (nali == 0 || status != eslEOF) eslx_msafile_ReadFailure(afp, status); /* a convenience, like eslx_msafile_OpenFailure() */
esl_alphabet_Destroy(abc);
eslx_msafile_Close(afp);
esl_randomness_Destroy(rng);
esl_getopts_Destroy(go);
exit(0);
}
/* multifetch:
* given a file containing lines with one name or key per line;
* parse the file line-by-line;
* if we have an SSI index available, retrieve the MSAs by key
* as we see each line;
* else, without an SSI index, store the keys in a hash, then
* read the entire MSA file in a single pass, outputting MSAs
* that are in our keylist.
*
* Note that with an SSI index, you get the MSAs in the order they
* appear in the <keyfile>, but without an SSI index, you get MSAs in
* the order they occur in the MSA file.
*/
static void
multifetch(ESL_GETOPTS *go, FILE *ofp, int outfmt, char *keyfile, ESLX_MSAFILE *afp)
{
ESL_KEYHASH *keys = esl_keyhash_Create();
ESL_FILEPARSER *efp = NULL;
ESL_MSA *msa = NULL;
int nali = 0;
char *key;
int keylen;
int keyidx;
int status;
if (esl_fileparser_Open(keyfile, NULL, &efp) != eslOK)
esl_fatal("Failed to open key file %s\n", keyfile);
esl_fileparser_SetCommentChar(efp, '#');
while (esl_fileparser_NextLine(efp) == eslOK)
{
if (esl_fileparser_GetTokenOnLine(efp, &key, &keylen) != eslOK)
esl_fatal("Failed to read MSA name on line %d of file %s\n", efp->linenumber, keyfile);
status = esl_keyhash_Store(keys, key, keylen, &keyidx);
if (status == eslEDUP) esl_fatal("MSA key %s occurs more than once in file %s\n", key, keyfile);
if (afp->ssi) { onefetch(go, ofp, outfmt, key, afp); nali++; }
}
if (! afp->ssi)
{
while ((status = eslx_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) eslx_msafile_ReadFailure(afp, status);
nali++;
if (msa->name == NULL)
esl_fatal("Every alignment in file must have a name to be retrievable. Failed to find name of alignment #%d\n", nali);
if ( (esl_keyhash_Lookup(keys, msa->name, -1, NULL) == eslOK) ||
(msa->acc != NULL && esl_keyhash_Lookup(keys, msa->acc, -1, NULL) == eslOK))
eslx_msafile_Write(ofp, msa, outfmt);
esl_msa_Destroy(msa);
}
}
if (ofp != stdout) printf("\nRetrieved %d alignments.\n", nali);
esl_keyhash_Destroy(keys);
esl_fileparser_Close(efp);
return;
}
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 */
ESLX_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 = eslx_msafile_Open(&abc, alifile, NULL, fmt, NULL, &afp)) != eslOK)
eslx_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 = eslx_msafile_Read(afp, &msa)) != eslEOF)
{
if (status != eslOK) eslx_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 = eslx_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);
}
eslx_msafile_Close(afp);
esl_getopts_Destroy(go);
return 0;
ERROR:
if(afp) eslx_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)
{
int i,j;
ESL_GETOPTS *go = NULL; /* command line processing */
ESL_STOPWATCH *w = esl_stopwatch_Create();
int status;
ESL_MSA *msa = NULL;
FILE *ofp = NULL; /* output file (default is stdout) */
ESL_ALPHABET *abc = NULL; /* digital alphabet */
char *alifile; /* name of the alignment file we're building HMMs from */
ESLX_MSAFILE *afp = NULL; /* open alifile */
int infmt = eslMSAFILE_UNKNOWN; /* autodetect alignment format by default. */
int outfmt = eslMSAFILE_STOCKHOLM;
char *postmsafile; /* optional file to resave annotated, modified MSAs to */
FILE *postmsafp = NULL; /* open <postmsafile>, or NULL */
int mask_range_cnt = 0;
uint32_t mask_starts[100]; // over-the-top allocation.
uint32_t mask_ends[100];
char *rangestr;
char *range;
int *map = NULL; /* map[i]=j, means model position i comes from column j of the alignment; 1..alen */
int keep_mm;
/* Set processor specific flags */
impl_Init();
alifile = NULL;
postmsafile = NULL;
/* Parse the command line
*/
process_commandline(argc, argv, &go, &alifile, &postmsafile);
keep_mm = esl_opt_IsUsed(go, "--apendmask");
/* Initialize what we can in the config structure (without knowing the alphabet yet).
* Fields controlled by masters are set up in usual_master() or mpi_master()
* Fields used by workers are set up in mpi_worker()
*/
ofp = NULL;
infmt = eslMSAFILE_UNKNOWN;
afp = NULL;
abc = NULL;
if (esl_opt_IsOn(go, "--informat")) {
infmt = eslx_msafile_EncodeFormat(esl_opt_GetString(go, "--informat"));
if (infmt == eslMSAFILE_UNKNOWN) p7_Fail("%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) p7_Fail(argv[0], "%s is not a recognized output MSA file format\n", esl_opt_GetString(go, "--outformat"));
/* Parse the ranges */
if (esl_opt_IsUsed(go, "--alirange")) {
esl_strdup(esl_opt_GetString(go, "--alirange"), -1, &rangestr) ;
} else if (esl_opt_IsUsed(go, "--modelrange")) {
esl_strdup(esl_opt_GetString(go, "--modelrange"), -1, &rangestr) ;
} else if (esl_opt_IsUsed(go, "--model2ali")) {
esl_strdup(esl_opt_GetString(go, "--model2ali"), -1, &rangestr) ;
} else if (esl_opt_IsUsed(go, "--ali2model")) {
esl_strdup(esl_opt_GetString(go, "--ali2model"), -1, &rangestr) ;
} else {
if (puts("Must specify mask range with --modelrange, --alirange, --model2ali, or --ali2model\n") < 0) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed"); goto ERROR;
}
while ( (status = esl_strtok(&rangestr, ",", &range) ) == eslOK) {
status = esl_regexp_ParseCoordString(range, mask_starts + mask_range_cnt, mask_ends + mask_range_cnt );
if (status == eslESYNTAX) esl_fatal("range flags take coords <from>..<to>; %s not recognized", range);
if (status == eslFAIL) esl_fatal("Failed to find <from> or <to> coord in %s", range);
mask_range_cnt++;
}
/* Start timing. */
esl_stopwatch_Start(w);
/* Open files, set alphabet.
* afp - open alignment file for input
* abc - alphabet expected or guessed in ali file
* postmsafp - open MSA output file
* ofp - optional open output file, or stdout
*/
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);
else abc = NULL;
status = eslx_msafile_Open(&abc, alifile, NULL, infmt, NULL, &afp);
if (status != eslOK) eslx_msafile_OpenFailure(afp, status);
if (esl_opt_IsUsed(go, "--alirange") || esl_opt_IsUsed(go, "--modelrange") ) {
postmsafp = fopen(postmsafile, "w");
if (postmsafp == NULL) p7_Fail("Failed to MSA output file %s for writing", postmsafile);
}
if (esl_opt_IsUsed(go, "-o"))
{
ofp = fopen(esl_opt_GetString(go, "-o"), "w");
if (ofp == NULL) p7_Fail("Failed to open -o output file %s\n", esl_opt_GetString(go, "-o"));
}
else ofp = stdout;
/* Looks like the i/o is set up successfully...
* Initial output to the user
*/
output_header(go, ofp, alifile, postmsafile); /* cheery output header */
/* read the alignment */
if ((status = eslx_msafile_Read(afp, &msa)) != eslOK) eslx_msafile_ReadFailure(afp, status);
if (esl_opt_IsUsed(go, "--alirange") || esl_opt_IsUsed(go, "--modelrange") ) {
/* add/modify mmline for the mask */
if (msa->mm == NULL) {
ESL_ALLOC(msa->mm, msa->alen);
keep_mm = FALSE;
}
if (!keep_mm)
for (i=0; i<msa->alen; i++) msa->mm[i] = '.';
}
// convert model coordinates to alignment coordinates, if necessary
if (esl_opt_IsUsed(go, "--modelrange") || esl_opt_IsUsed(go, "--model2ali") || esl_opt_IsUsed(go, "--ali2model") ) {
float symfrac = esl_opt_GetReal(go, "--symfrac");
int do_hand = esl_opt_IsOn(go, "--hand");
int L;
//same as p7_builder relative_weights
if (esl_opt_IsOn(go, "--wnone") ) { esl_vec_DSet(msa->wgt, msa->nseq, 1.); }
else if (esl_opt_IsOn(go, "--wgiven") ) ;
else if (esl_opt_IsOn(go, "--wpb") ) status = esl_msaweight_PB(msa);
else if (esl_opt_IsOn(go, "--wgsc") ) status = esl_msaweight_GSC(msa);
else if (esl_opt_IsOn(go, "--wblosum")) status = esl_msaweight_BLOSUM(msa, esl_opt_GetReal(go, "--wid"));
if ((status = esl_msa_MarkFragments(msa, esl_opt_GetReal(go, "--fragthresh"))) != eslOK) goto ERROR;
//build a map of model mask coordinates to alignment coords
ESL_ALLOC(map, sizeof(int) * (msa->alen+1));
L = p7_Alimask_MakeModel2AliMap(msa, do_hand, symfrac, map );
if ( esl_opt_IsUsed(go, "--model2ali") ) {
//print mapping
printf ("model coordinates alignment coordinates\n");
for (i=0; i<mask_range_cnt; i++)
printf ("%8d..%-8d -> %8d..%-8d\n", mask_starts[i], mask_ends[i], map[mask_starts[i]-1], map[mask_ends[i]-1]);
/* If I wanted to, I could print all the map values independently:
printf("\n\n-----------\n");
printf("Map\n");
printf("---\n");
for (i=0; i<L; i++)
printf("%d -> %d\n", i+1, map[i]);
*/
} else if ( esl_opt_IsUsed(go, "--ali2model") ) {
//print mapping (requires scanning the inverted map
int alistart = 0;
int aliend = 0;
printf ("alignment coordinates model coordinates\n");
for (i=0; i<mask_range_cnt; i++) {
//find j for ali positions
while (map[alistart] < mask_starts[i] )
alistart++;
aliend = alistart;
while (map[aliend] < mask_ends[i] )
aliend++;
printf (" %8d..%-8d -> %8d..%-8d\n", map[alistart], map[aliend], alistart+1, aliend+1);
}
} else {
//convert the mask coords based on map
for (i=0; i<mask_range_cnt; i++) {
mask_starts[i] = map[mask_starts[i]-1]; //-1 because mmline is offset by one relative to the 1-base alignment
mask_ends[i] = map[mask_ends[i]-1];
}
}
}
if (esl_opt_IsUsed(go, "--alirange") || esl_opt_IsUsed(go, "--modelrange") ) {
//overwrite '.' with 'm' everywhere the range says to do it
for (i=0; i<mask_range_cnt; i++)
for (j=mask_starts[i]; j<=mask_ends[i]; j++)
msa->mm[j-1] = 'm';
if ((status = eslx_msafile_Write(postmsafp, msa, outfmt)) != eslOK) ESL_XEXCEPTION_SYS(eslEWRITE, "write failed");
}
esl_stopwatch_Stop(w);
if (esl_opt_IsOn(go, "-o")) fclose(ofp);
if (postmsafp) fclose(postmsafp);
if (afp) eslx_msafile_Close(afp);
if (abc) esl_alphabet_Destroy(abc);
esl_getopts_Destroy(go);
esl_stopwatch_Destroy(w);
return 0;
ERROR:
return eslFAIL;
}
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; /* 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;
}
