void
MSAFileWrite(FILE *fp, MSA *msa, int outfmt, int do_oneline)
{
switch (outfmt) {
case MSAFILE_A2M: WriteA2M(fp, msa); break;
case MSAFILE_CLUSTAL: WriteClustal(fp, msa); break;
case MSAFILE_MSF: WriteMSF(fp, msa); break;
case MSAFILE_PHYLIP: WritePhylip(fp, msa); break;
case MSAFILE_SELEX: WriteSELEX(fp, msa); break;
case MSAFILE_STOCKHOLM:
if (do_oneline) WriteStockholmOneBlock(fp, msa);
else WriteStockholm(fp, msa);
break;
default:
Die("can't write. no such alignment format %d\n", outfmt);
}
}
/*****************************************************************
* msf.c test driver:
* cc -DTESTDRIVE_MSF -g -O2 -Wall -o test msf.c msa.c gki.c sqerror.c sre_string.c file.c hsregex.c sre_math.c sre_ctype.c sqio.c alignio.c selex.c interleaved.c types.c -lm
*
*/
int
main(int argc, char **argv)
{
MSAFILE *afp;
MSA *msa;
char *file;
file = argv[1];
if ((afp = MSAFileOpen(file, MSAFILE_STOCKHOLM, NULL)) == NULL)
Die("Couldn't open %s\n", file);
while ((msa = ReadMSF(afp)) != NULL)
{
WriteMSF(stdout, msa);
MSAFree(msa);
}
MSAFileClose(afp);
exit(0);
}
/**
* @brief Write alignment to file.
*
* @param[in] mseq
* The mseq_t struct containing the aligned sequences
* @param[in] pcAlnOutfile
* The name of the output file
* @param[in] outfmt
* The alignment output format (defined in squid.h)
* @param[in] iWrap
* length of line for Clustal/Fasta format
*
* @return Non-zero on error
*
* @note We create a temporary squid MSA struct in here because we never
* use it within clustal. We might be better of using the old clustal
* output routines instead.
*
*/
int
WriteAlignment(mseq_t *mseq, const char *pcAlnOutfile, int outfmt, int iWrap, bool bResno)
{
int i; /* aux */
MSA *msa; /* squid's alignment structure */
FILE *pfOut = NULL;
int key; /* MSA struct internal index for sequence */
int alen; /* alignment length */
bool use_stdout;
assert(mseq!=NULL);
if (MSAFILE_UNKNOWN == outfmt) {
Log(&rLog, LOG_ERROR, "Unknown output format chosen");
return -1;
}
if (NULL == pcAlnOutfile) {
pfOut = stdout;
use_stdout = TRUE;
} else {
use_stdout = FALSE;
if (NULL == (pfOut = fopen(pcAlnOutfile, "w"))) {
Log(&rLog, LOG_ERROR, "Could not open file %s for writing", pcAlnOutfile);
return -1;
}
}
/* derive alignment length from first seq */
alen = strlen(mseq->seq[0]);
msa = MSAAlloc(mseq->nseqs, alen);
/* basic structure borrowed code from squid-1.9g/a2m.c:ReadA2M()
* we actually create a copy of mseq. keeping the pointers becomes
* messy when calling MSAFree()
*/
for (i=0; i<mseq->nseqs; i++) {
char *this_name = NULL; /* mseq sequence name */
char *this_seq = NULL; /* mseq sequence */
SQINFO *this_sqinfo = NULL; /* mseq sequence name */
int iI;
/* mseq->tree_order encodes to order in which sequences are listed in the guide-tree,
if the user wants the sequence output in the input-order then mseq->tree_order==NULL,
otherwise mseq->tree_order!=NULL, containing the indices of the sequences, FS, r274 -> */
iI = (NULL == mseq->tree_order) ? i : mseq->tree_order[i];
this_name = mseq->sqinfo[iI].name; /* mseq sequence name */
this_seq = mseq->seq[iI]; /* mseq sequence */
this_sqinfo = &mseq->sqinfo[iI]; /* mseq sequence name */
key = GKIStoreKey(msa->index, this_name);
msa->sqname[key] = sre_strdup(this_name, strlen(this_name));
/* setting msa->sqlen[idx] and msa->aseq[idx] */
msa->sqlen[key] = sre_strcat(&(msa->aseq[key]), msa->sqlen[key],
this_seq, strlen(this_seq));
if (this_sqinfo->flags & SQINFO_DESC) {
/* FIXME never get here ... */
MSASetSeqDescription(msa, key, this_sqinfo->desc);
}
/* FIXME extend this by copying more stuff according to flags.
* See MSAFileRead() in msa.c and used functions there
*
* Problem is that we never parse MSA information as we use squid'sSeqFile
*/
msa->nseq++;
} /* 0 <= i < mseq->nseqs */
/* FIXME Would like to, but can't use MSAVerifyParse(msa) here, as it
* will die on error. Need to implement our own version
*/
#if 0
MSAVerifyParse(msa);
#endif
/* The below is copy of MSAFileWrite() which originally only writes to stdout.
*/
/* Be sloppy and make a2m and fasta the same. same for vienna (which is
the same). same same. can can. boleh boleh */
if (outfmt==SQFILE_FASTA)
outfmt = MSAFILE_A2M;
if (outfmt==SQFILE_VIENNA)
outfmt = MSAFILE_VIENNA;
switch (outfmt) {
case MSAFILE_A2M:
/*WriteA2M(pfOut, msa, 0);*/
WriteA2M(pfOut, msa, iWrap);
break;
case MSAFILE_VIENNA:
/*WriteA2M(pfOut, msa, 1);*/
WriteA2M(pfOut, msa, INT_MAX);
break;
case MSAFILE_CLUSTAL:
WriteClustal(pfOut, msa, iWrap, TRUE==bResno ? 1 : 0, mseq->seqtype);
break;
case MSAFILE_MSF:
WriteMSF(pfOut, msa);
break;
case MSAFILE_PHYLIP:
WritePhylip(pfOut, msa);
break;
case MSAFILE_SELEX:
WriteSELEX(pfOut, msa);
break;
case MSAFILE_STOCKHOLM:
WriteStockholm(pfOut, msa);
break;
default:
Log(&rLog, LOG_FATAL, "internal error: %s",
"invalid output format should have been detected before");
}
if (use_stdout == FALSE) {
(void) fclose(pfOut);
Log(&rLog, LOG_INFO,
"Alignment written to %s", pcAlnOutfile);
}
MSAFree(msa);
return 0;
}
