/* 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;
}
