/* Function: ReadMultipleRseqs()
*
* Purpose: Open a data file and
* parse it into an array of rseqs (raw, unaligned
* sequences).
*
* Caller is responsible for free'ing memory allocated
* to ret_rseqs, ret_weights, and ret_names.
*
* Weights are currently only supported for MSF format.
* Sequences read from all other formats will be assigned
* weights of 1.0. If the caller isn't interested in
* weights, it passes NULL as ret_weights.
*
* Returns 1 on success. Returns 0 on failure and sets
* squid_errno to indicate the cause.
*/
int
ReadMultipleRseqs(char *seqfile,
int fformat,
char ***ret_rseqs,
SQINFO **ret_sqinfo,
int *ret_num)
{
SQINFO *sqinfo; /* array of sequence optional info */
SQFILE *dbfp; /* open ptr for sequential access of file */
char **rseqs; /* sequence array */
char **aseqs; /* aligned sequences, if file is aligned */
AINFO ainfo; /* alignment-associated information */
int numalloced; /* num of seqs currently alloced for */
int idx;
int num;
if (fformat == kSelex || fformat == kMSF || fformat == kClustal)
{
if (! ReadAlignment(seqfile, fformat, &aseqs, &ainfo)) return 0;
if (! DealignAseqs(aseqs, ainfo.nseq, &rseqs)) return 0;
/* copy the sqinfo array
*/
num = ainfo.nseq;
sqinfo= (SQINFO *) MallocOrDie (sizeof(SQINFO)*ainfo.nseq);
for (idx = 0; idx < ainfo.nseq; idx++)
SeqinfoCopy(&(sqinfo[idx]), &(ainfo.sqinfo[idx]));
FreeAlignment(aseqs, &ainfo);
}
else
{
/* initial alloc */
num = 0;
numalloced = 16;
rseqs = (char **) MallocOrDie (numalloced * sizeof(char *));
sqinfo = (SQINFO *) MallocOrDie (numalloced * sizeof(SQINFO));
if ((dbfp = SeqfileOpen(seqfile, fformat, NULL)) == NULL) return 0;
while (ReadSeq(dbfp, fformat, &rseqs[num], &(sqinfo[num])))
{
num++;
if (num == numalloced) /* more seqs coming, alloc more room */
{
numalloced += 16;
rseqs = (char **) ReallocOrDie (rseqs, numalloced*sizeof(char *));
sqinfo = (SQINFO *) ReallocOrDie (sqinfo, numalloced * sizeof(SQINFO));
}
}
SeqfileClose(dbfp);
}
*ret_rseqs = rseqs;
*ret_sqinfo = sqinfo;
*ret_num = num;
return 1;
}
/**
* @brief reads sequences from file
*
* @param[out] prMSeq
* Multiple sequence struct. Must be preallocated.
* FIXME: would make more sense to allocate it here.
* @param[in] seqfile
* Sequence file name. If '-' sequence will be read from stdin.
* @param[in] iSeqType
* int-encoded sequence type. Set to
* SEQTYPE_UNKNOWN for autodetect (guessed from first sequence)
* @param[in] iMaxNumSeq
* Return an error, if more than iMaxNumSeq have been read
* @param[in] iMaxSeqLen
* Return an error, if a seq longer than iMaxSeqLen has been read
*
* @return 0 on success, -1 on error
*
* @note
* - Depends heavily on squid
* - Sequence file format will be guessed
* - If supported by squid, gzipped files can be read as well.
*/
int
ReadSequences(mseq_t *prMSeq, char *seqfile,
int iSeqType, int iSeqFmt, bool bIsProfile, bool bDealignInputSeqs,
int iMaxNumSeq, int iMaxSeqLen)
{
SQFILE *dbfp; /* sequence file descriptor */
char *cur_seq;
SQINFO cur_sqinfo;
int iSeqIdx; /* sequence counter */
int iSeqPos; /* sequence string position counter */
assert(NULL!=seqfile);
/* Try to work around inability to autodetect from a pipe or .gz:
* assume FASTA format
*/
if (SQFILE_UNKNOWN == iSeqFmt &&
(Strparse("^.*\\.gz$", seqfile, 0) || strcmp(seqfile, "-") == 0)) {
iSeqFmt = SQFILE_FASTA;
}
/* Using squid routines to read input. taken from seqstat_main.c. we don't
* know if input is aligned, so we use SeqfileOpen instead of MSAFileOpen
* etc. NOTE this also means we discard some information, e.g. when
* reading from and writing to a stockholm file, all extra MSA
* info/annotation will be lost.
*
*/
if (NULL == (dbfp = SeqfileOpen(seqfile, iSeqFmt, NULL))) {
Log(&rLog, LOG_ERROR, "Failed to open sequence file %s for reading", seqfile);
return -1;
}
/* FIXME squid's ReadSeq() will exit with fatal error if format is
* unknown. This will be a problem for a GUI. Same is true for many squid
* other functions.
*
* The original squid:ReadSeq() dealigns sequences on input. We
* use a patched version.
*
*/
while (ReadSeq(dbfp, dbfp->format,
&cur_seq,
&cur_sqinfo)) {
if (prMSeq->nseqs+1>iMaxNumSeq) {
Log(&rLog, LOG_ERROR, "Maximum number of sequences (=%d) exceeded after reading sequence '%s' from '%s'",
iMaxNumSeq, cur_sqinfo.name, seqfile);
return -1;
}
if ((int)strlen(cur_seq)>iMaxSeqLen) {
Log(&rLog, LOG_ERROR, "Sequence '%s' has %d residues and is therefore longer than allowed (max. sequence length is %d)",
cur_sqinfo.name, strlen(cur_seq), iMaxSeqLen);
return -1;
}
if ((int)strlen(cur_seq)==0) {
Log(&rLog, LOG_ERROR, "Sequence '%s' has 0 residues",
cur_sqinfo.name);
return -1;
}
/* FIXME: use modified version of AddSeq() that allows handing down SqInfo
*/
prMSeq->seq = (char **)
CKREALLOC(prMSeq->seq, (prMSeq->nseqs+1) * sizeof(char *));
prMSeq->seq[prMSeq->nseqs] = CkStrdup(cur_seq);
prMSeq->sqinfo = (SQINFO *)
CKREALLOC(prMSeq->sqinfo, (prMSeq->nseqs+1) * sizeof(SQINFO));
SeqinfoCopy(&prMSeq->sqinfo[prMSeq->nseqs], &cur_sqinfo);
#ifdef TRACE
Log(&rLog, LOG_FORCED_DEBUG, "seq no %d: seq = %s", prMSeq->nseqs, prMSeq->seq[prMSeq->nseqs]);
LogSqInfo(&prMSeq->sqinfo[prMSeq->nseqs]);
#endif
/* always guess type from first seq. use squid function and
* convert value
*/
if (0 == prMSeq->nseqs) {
int type = Seqtype(prMSeq->seq[prMSeq->nseqs]);
switch (type) {
case kDNA:
prMSeq->seqtype = SEQTYPE_DNA;
break;
case kRNA:
prMSeq->seqtype = SEQTYPE_RNA;
break;
case kAmino:
prMSeq->seqtype = SEQTYPE_PROTEIN;
break;
case kOtherSeq:
prMSeq->seqtype = SEQTYPE_UNKNOWN;
break;
default:
Log(&rLog, LOG_FATAL, "Internal error in %s", __FUNCTION__);
}
/* override with given sequence type but check with
* automatically detected type and warn if necessary
*/
if (SEQTYPE_UNKNOWN != iSeqType) {
if (prMSeq->seqtype != iSeqType) {
Log(&rLog, LOG_WARN, "Overriding automatically determined seq-type %s to %s as requested",
SeqTypeToStr(prMSeq->seqtype), SeqTypeToStr(iSeqType));
prMSeq->seqtype = iSeqType;
}
}
/* if type could not be determined and was not set return error */
if (SEQTYPE_UNKNOWN == iSeqType && SEQTYPE_UNKNOWN == prMSeq->seqtype) {
Log(&rLog, LOG_ERROR, "Couldn't guess sequence type from first sequence");
FreeSequence(cur_seq, &cur_sqinfo);
SeqfileClose(dbfp);
return -1;
}
}
Log(&rLog, LOG_DEBUG, "seq-no %d: type=%s name=%s len=%d seq=%s",
prMSeq->nseqs, SeqTypeToStr(prMSeq->seqtype),
prMSeq->sqinfo[prMSeq->nseqs].name, prMSeq->sqinfo[prMSeq->nseqs].len,
prMSeq->seq[prMSeq->nseqs]);
/* FIXME IPUAC and/or case conversion? If yes see
* corresponding squid functions. Special treatment of
* Stockholm tilde-gaps for ktuple code?
*/
prMSeq->nseqs++;
FreeSequence(cur_seq, &cur_sqinfo);
}
SeqfileClose(dbfp);
/*#if ALLOW_ONLY_PROTEIN
if (SEQTYPE_PROTEIN != prMSeq->seqtype) {
Log(&rLog, LOG_FATAL, "Sequence type is %s. %s only works on protein.",
SeqTypeToStr(prMSeq->seqtype), PACKAGE_NAME);
}
#endif*/
/* Check if sequences are aligned */
prMSeq->aligned = SeqsAreAligned(prMSeq, bIsProfile, bDealignInputSeqs);
/* keep original sequence as copy and convert "working" sequence
*
*/
prMSeq->orig_seq = (char**) CKMALLOC(prMSeq->nseqs * sizeof(char *));
for (iSeqIdx=0; iSeqIdx<prMSeq->nseqs; iSeqIdx++) {
prMSeq->orig_seq[iSeqIdx] = CkStrdup(prMSeq->seq[iSeqIdx]);
/* convert unknown characters according to set seqtype
* be conservative, i.e. don't allow any fancy ambiguity
* characters to make sure that ktuple code etc. works.
*/
/* first on the fly conversion between DNA and RNA
*/
if (prMSeq->seqtype==SEQTYPE_DNA)
ToDNA(prMSeq->seq[iSeqIdx]);
if (prMSeq->seqtype==SEQTYPE_RNA)
ToRNA(prMSeq->seq[iSeqIdx]);
/* then check of each character
*/
for (iSeqPos=0; iSeqPos<(int)strlen(prMSeq->seq[iSeqIdx]); iSeqPos++) {
char *res = &(prMSeq->seq[iSeqIdx][iSeqPos]);
if (isgap(*res))
continue;
if (prMSeq->seqtype==SEQTYPE_PROTEIN) {
if (NULL == strchr(AMINO_ALPHABET, toupper(*res))) {
*res = AMINOACID_ANY;
}
} else if (prMSeq->seqtype==SEQTYPE_DNA) {
if (NULL == strchr(DNA_ALPHABET, toupper(*res))) {
*res = NUCLEOTIDE_ANY;
}
} else if (prMSeq->seqtype==SEQTYPE_RNA) {
if (NULL == strchr(RNA_ALPHABET, toupper(*res))) {
*res = NUCLEOTIDE_ANY;
}
}
}
}
/* order in which sequences appear in guide-tree
* only allocate if different output-order desired */
prMSeq->tree_order = NULL;
prMSeq->filename = CkStrdup(seqfile);
Log(&rLog, LOG_INFO, "Read %d sequences (type: %s) from %s",
prMSeq->nseqs, SeqTypeToStr(prMSeq->seqtype), prMSeq->filename);
return 0;
}
main (int argc, char ** argv )
{
char *seqfile; /* name of sequence file */
SQINFO sqinfo; /* extra info about sequence */
SQFILE *dbfp; /* open sequence file */
int fmt,ofmt=106; /* format of seqfile */
/* 106 is PHYLIP format in SQUID */
char *seq; /* sequence */
int type; /* kAmino, kDNA, kRNA, or kOtherSeq */
sequence * seqs, * cds_seqs;
sequence tmp_seqs[2], tmp_cds_seqs[2];
char *optname;
char *optarg, *t;
int optind;
int be_quiet;
int seqct = 0,cdsct = 0;
int min_aln_len = 0;
int do_oneline = 0;
char * output_filename = 0, *submat_file = 0;
int showaln = 1;
int showheader=1;
FILE *ofd, *fd;
alignment *cds_aln;
alignment * opt_alignment = NULL; /* place for pairwise alignment */
int len,i,j, k, jk,ik,aln_count, rc;
pairwise_distances pwMLdist, pwNGdist;
int firsttime = 1;
struct timeval tp;
pwMLdist.N = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwMLdist.dN = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwMLdist.S = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwMLdist.dS = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwMLdist.dNdS = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwMLdist.SEdS = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwMLdist.SEdN = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwMLdist.t = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwMLdist.kappa= make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwNGdist.dN = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwNGdist.dS = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
pwNGdist.dNdS = make_double_matrix(NUM_PW_SEQS,NUM_PW_SEQS);
/*
pwMLdist.N = pwMLdist.dN = pwMLdist.S = 0;
pwMLdist.dS = pwMLdist.dNdS = pwMLdist.SEdS = 0;
pwMLdist.SEdN = pwMLdist.t = pwMLdist.kappa= 0;
pwNGdist.dN = pwNGdist.dS = pwNGdist.dNdS = 0;
*/
Alntype = default_aln_type;
/* Command line Parse */
fmt = SQFILE_UNKNOWN; /* default: autodetect format */
be_quiet = FALSE;
type = kOtherSeq;
/* for our purposes this is only pairwise alignments, but
* would rather do it correctly in case we move to MSA case
*/
while (Getopt(argc, argv, OPTIONS, NOPTIONS, usage,
&optind, &optname, &optarg))
{
if (strcmp(optname, "--matrix") == 0) submat_file = optarg;
else if (strcmp(optname, "--quiet") == 0) be_quiet = TRUE;
else if (strcmp(optname, "--gapopen") == 0) {
Gapopen = atoi(optarg);
if( Gapopen < 0 ) Gapopen *= -1;
} else if (strcmp(optname, "--gapext") == 0) {
Gapext = atoi(optarg);
if( Gapext < 0 ) Gapext *= -1;
} else if (strcmp(optname, "--informat") == 0) {
fmt = String2SeqfileFormat(optarg);
if (fmt == SQFILE_UNKNOWN)
Die("unrecognized sequence file format \"%s\"", optarg);
} else if (strcmp(optname, "--outformat") == 0) {
ofmt = String2SeqfileFormat(optarg);
if (ofmt == SQFILE_UNKNOWN)
Die("unrecognized sequence file format \"%s\"", optarg);
} else if( strcmp(optname, "--global") == 0 ) {
Alntype = global;
} else if (strcmp(optname, "-h") == 0) {
puts(usage);
puts(experts);
exit(EXIT_SUCCESS);
} else if ( strcmp(optname, "-v") == 0 ) {
Verbose = 1;
} else if ( strcmp(optname, "--gapchar") == 0 ) {
GapChar = optarg[0];
} else if( strcmp(optname, "--output") == 0 ) {
output_filename = optarg;
} else if( strcmp(optname, "--showtable" ) == 0 ) {
showaln = 0;
} else if( strcmp(optname, "--noheader" ) == 0 ) {
showheader = 0;
}
}
if (argc - optind < 1) Die("%s\n", usage);
if( ! submat_file ) {
if( (t = getenv("SUBOPTDIR")) != 0 ||
(t = getenv("SUBOPT_DIR")) != 0 ) {
submat_file = calloc(strlen(t) + 24, sizeof(char));
sprintf(submat_file, "%s/%s",t,Default_submat);
} else {
submat_file = calloc(strlen((void *)Default_submat) + 24, sizeof(char));
sprintf(submat_file, "../%s",Default_submat);
}
}
/* open matrix */
fd = fopen(submat_file, "r");
if( ! ParsePAMFile(fd,&ScoringMatrix, &MatrixScale) ) {
fprintf(stderr, "Cannot parse or open matrix file %s\n",submat_file);
free(submat_file);
exit(EXIT_SUCCESS);
}
if( output_filename && strlen(output_filename) != 1 &&
output_filename[0] != '-') {
ofd = fopen(output_filename,"w");
if( ! ofd ) {
fprintf(stderr, "could not open file %s",output_filename);
goto end;
}
} else
ofd = stdout;
while( optind < argc ) {
seqfile = argv[optind++];
/* Try to work around inability to autodetect from a pipe or .gz:
* assume FASTA format
*/
if (fmt == SQFILE_UNKNOWN &&
(Strparse("^.*\\.gz$", seqfile, 0) || strcmp(seqfile, "-") == 0))
fmt = SQFILE_FASTA;
if ((dbfp = SeqfileOpen(seqfile, fmt, NULL)) == NULL)
Die("Failed to open sequence file %s for reading", seqfile);
while (ReadSeq(dbfp, dbfp->format, &seq, &sqinfo))
{
FreeSequence(NULL, &sqinfo);
seqct++;
}
cds_seqs = (sequence *)calloc(seqct, sizeof(sequence));
seqs = (sequence *)calloc(seqct, sizeof(sequence));
SeqfileRewind(dbfp);
seqct=0;
while (ReadSeq(dbfp, dbfp->format, &seq, &sqinfo))
{
sqinfo.type = Seqtype(seq);
if( sqinfo.type == kDNA || sqinfo.type == kRNA ) {
seqs[seqct].seqstr = Translate(seq,stdcode1);
/* Let's remove the last codon if it is a stop codon */
len = strlen(seqs[seqct].seqstr);
if( Verbose )
fprintf(stderr,"seqct is %d length is %d\n",seqct,
len);
if( seqs[seqct].seqstr[len-1] == '*' ) {
seqs[seqct].seqstr[len-1] = '\0';
seq[strlen(seq) - 3] = '\0';
}
cds_seqs[cdsct].seqstr = seq;
seqs[seqct].seqname = calloc(strlen(sqinfo.name)+1,sizeof(char));
cds_seqs[cdsct].seqname = calloc(strlen(sqinfo.name)+1,sizeof(char));
strcpy(seqs[seqct].seqname,sqinfo.name );
strcpy(cds_seqs[cdsct].seqname,sqinfo.name);
cds_seqs[cdsct].length = sqinfo.len;
cds_seqs[cdsct].alphabet = ( sqinfo.type == kDNA ) ? dna : rna;
seqs[seqct].length = strlen(seqs[seqct].seqstr);
seqs[seqct].alphabet = protein;
cdsct++; seqct++;
} else {
fprintf(stderr,"Expect CDS sequences (DNA or RNA) not Protein\n");
goto end;
}
FreeSequence(NULL, &sqinfo);
if( Verbose && seqct > 3 )
break;
}
if( seqct < 2 ) {
fprintf(stderr,"Must have provided a valid file with at least 2 sequences in it");
goto end;
}
for( i=0; i < seqct; i++ ) {
for(k=i+1; k < seqct; k++ ) {
if( (opt_alignment = (alignment *)calloc(1,sizeof(alignment *))) == NULL) {
fprintf(stderr,"Could not allocate memory\n");
goto end;
}
opt_alignment->msa = NULL;
rc = optimal_align(&seqs[i],&seqs[k],opt_alignment);
if( rc != 1 ) {
fprintf(stderr,"Could not make an optimal alignment\n");
goto end;
} else {
tmp_cds_seqs[0] = cds_seqs[i];
tmp_cds_seqs[1] = cds_seqs[k];
rc = mrtrans(opt_alignment, tmp_cds_seqs, &cds_aln,0);
if( rc != 0 ) {
fprintf(stderr, "Could not map the coding sequence to the protein alignemnt for aln %d: %d\n",i,rc);
goto end;
}
if( showaln ) {
if( ofmt >= 100 ) {
MSAFileWrite(ofd,cds_aln->msa, ofmt,do_oneline);
} else {
for(j=0; j < cds_aln->msa->nseq; j++ ) {
WriteSeq(ofd, ofmt,
cds_aln->msa->aseq[j],
&(cds_aln->sqinfo[j]) );
}
}
} else {
if( showheader && firsttime ) {
fprintf(ofd,"SEQ1\tSEQ2\tSCORE\tdN\tdS\tOMEGA\tN\tS\tkappa\tt\tLENGTH\n");
firsttime = 0;
}
if( do_kaks_yn00(cds_aln->msa, &pwMLdist,&pwNGdist) < 0 ) {
fprintf(stderr, "warning: problem with align for %s %s\n",
cds_aln->msa->sqname[0], cds_aln->msa->sqname[1]);
continue;
}
for(ik = 0; ik < NUM_PW_SEQS; ik++ ) {
for( jk = ik+1; jk < NUM_PW_SEQS; jk++ ) {
fprintf(ofd,"%s\t%s\t%d\t%f\t%f\t%f\t%f\t%f\t%f\t%f\t%d\n",
cds_aln->sqinfo[ik].name,
cds_aln->sqinfo[jk].name,
opt_alignment->score,
pwMLdist.dN[ik][jk],pwMLdist.dS[ik][jk],
pwMLdist.dNdS[ik][jk],
pwMLdist.N[ik][jk],
pwMLdist.S[ik][jk],
pwMLdist.kappa[ik][jk],
pwMLdist.t[ik][jk],
opt_alignment->msa->alen);
}
}
}
}
cleanup_alignment(cds_aln);
cleanup_alignment(opt_alignment);
}
}
}
if( ofd && ofd != stdout )
fclose(ofd);
end:
free(submat_file);
Free2DArray((void **)ScoringMatrix,27);
for(i =0; i< seqct; i++ ) {
free(seqs[i].seqstr);
free(seqs[i].seqname);
seqs[i].seqstr = seqs[i].seqname = 0;
}
for(i = 0; i < cdsct; i++) {
free(cds_seqs[i].seqstr);
free(cds_seqs[i].seqname);
cds_seqs[i].seqstr = cds_seqs[i].seqname = 0;
}
cleanup_matrix((void **)pwMLdist.N,NUM_PW_SEQS);
cleanup_matrix((void **)pwMLdist.dN,NUM_PW_SEQS);
cleanup_matrix((void **)pwMLdist.S,NUM_PW_SEQS);
cleanup_matrix((void **)pwMLdist.dS,NUM_PW_SEQS);
cleanup_matrix((void **)pwMLdist.SEdS,NUM_PW_SEQS);
cleanup_matrix((void **)pwMLdist.SEdN,NUM_PW_SEQS);
cleanup_matrix((void **)pwMLdist.t,NUM_PW_SEQS);
cleanup_matrix((void **)pwMLdist.dNdS,NUM_PW_SEQS);
cleanup_matrix((void **)pwMLdist.kappa,NUM_PW_SEQS);
cleanup_matrix((void **)pwNGdist.dN,NUM_PW_SEQS);
cleanup_matrix((void **)pwNGdist.dS,NUM_PW_SEQS);
cleanup_matrix((void **)pwNGdist.dNdS,NUM_PW_SEQS);
free(pwNGdist.dNdS);
free(pwNGdist.dN);
free(pwNGdist.dS);
free(pwMLdist.dNdS);
free(pwMLdist.dN);
free(pwMLdist.dS);
free(pwMLdist.N);
free(pwMLdist.S);
free(pwMLdist.SEdS);
free(pwMLdist.SEdN);
free(pwMLdist.t);
free(pwMLdist.kappa);
return 0;
}
