static void
onefetch_subseq(ESL_GETOPTS *go, FILE *ofp, ESL_SQFILE *sqfp, char *newname, char *key, uint32_t given_start, uint32_t given_end)
{
int start, end;
int do_revcomp;
ESL_SQ *sq = esl_sq_Create();
if (sqfp->data.ascii.ssi == NULL) esl_fatal("no ssi index");
/* reverse complement indicated by coords. */
/* -c 52: would be 52,0, so watch out for given_end = 0 case */
if (given_end != 0 && given_start > given_end)
{ start = given_end; end = given_start; do_revcomp = TRUE; }
else
{ start = given_start; end = given_end; do_revcomp = FALSE; }
if (esl_sqio_FetchSubseq(sqfp, key, start, end, sq) != eslOK) esl_fatal(esl_sqfile_GetErrorBuf(sqfp));
if (newname != NULL) esl_sq_SetName(sq, newname);
else esl_sq_FormatName(sq, "%s/%d-%d", key, given_start, (given_end == 0) ? sq->L : given_end);
/* Two ways we might have been asked to revcomp: by coord, or by -r option */
/* (If both happen, they'll cancel each other out) */
if (do_revcomp)
if (esl_sq_ReverseComplement(sq) != eslOK) esl_fatal("Failed to reverse complement %s; is it a protein?\n", sq->name);
if (esl_opt_GetBoolean(go, "-r"))
if (esl_sq_ReverseComplement(sq) != eslOK) esl_fatal("Failed to reverse complement %s; is it a protein?\n", sq->name);
esl_sqio_Write(ofp, sq, eslSQFILE_FASTA, FALSE);
esl_sq_Destroy(sq);
}
static int
do_by_sequences(ESL_GENCODE *gcode, ESL_GENCODE_WORKSTATE *wrk, ESL_SQFILE *sqfp)
{
ESL_SQ *sq = esl_sq_CreateDigital(gcode->nt_abc);
int status;
while (( status = esl_sqio_Read(sqfp, sq )) == eslOK)
{
if (sq->n < 3) continue;
if (wrk->do_watson) {
esl_gencode_ProcessStart(gcode, wrk, sq);
esl_gencode_ProcessPiece(gcode, wrk, sq);
esl_gencode_ProcessEnd(wrk, sq);
}
if (wrk->do_crick) {
esl_sq_ReverseComplement(sq);
esl_gencode_ProcessStart(gcode, wrk, sq);
esl_gencode_ProcessPiece(gcode, wrk, sq);
esl_gencode_ProcessEnd(wrk, sq);
}
esl_sq_Reuse(sq);
}
if (status == eslEFORMAT) esl_fatal("Parse failed (sequence file %s)\n%s\n",
sqfp->filename, sqfp->get_error(sqfp));
else if (status != eslEOF) esl_fatal("Unexpected error %d reading sequence file %s",
status, sqfp->filename);
esl_sq_Destroy(sq);
return eslOK;
}
/* onefetch():
* Given one <key> (a seq name or accession), retrieve the corresponding sequence.
* In SSI mode, we can do this quickly by positioning the file, then regurgitating
* every line until the end-of-record marker; we don't even have to parse.
* Without an SSI index, we have to parse the file sequentially 'til we find
* the one we're after.
*/
static void
onefetch(ESL_GETOPTS *go, FILE *ofp, char *key, ESL_SQFILE *sqfp)
{
ESL_SQ *sq = esl_sq_Create();
int do_revcomp = esl_opt_GetBoolean(go, "-r");
char *newname = esl_opt_GetString(go, "-n");
int status;
/* Try to position the file at the desired sequence with SSI. */
if (sqfp->data.ascii.ssi != NULL)
{
status = esl_sqfile_PositionByKey(sqfp, key);
if (status == eslENOTFOUND) esl_fatal("seq %s not found in SSI index for file %s\n", key, sqfp->filename);
else if (status == eslEFORMAT) esl_fatal("Failed to parse SSI index for %s\n", sqfp->filename);
else if (status != eslOK) esl_fatal("Failed to look up location of seq %s in SSI index of file %s\n", key, sqfp->filename);
status = esl_sqio_Read(sqfp, sq);
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 EOF reading sequence file %s",
status, sqfp->filename);
else if (status != eslOK) esl_fatal("Unexpected error %d reading sequence file %s",
status, sqfp->filename);
if (strcmp(key, sq->name) != 0 && strcmp(key, sq->acc) != 0)
esl_fatal("whoa, internal error; found the wrong sequence %s, not %s", sq->name, key);
}
else
{ /* Else, we have to read the whole damn file sequentially until we find the seq */
while ((status = esl_sqio_Read(sqfp, sq)) != eslEOF) {
if (status == eslEFORMAT) esl_fatal("Parse failed (sequence file %s):\n%s\n",
sqfp->filename, esl_sqfile_GetErrorBuf(sqfp));
else if (status != eslOK) esl_fatal("Unexpected error %d reading sequence file %s",
status, sqfp->filename);
if (strcmp(key, sq->name) == 0 || strcmp(key, sq->acc) == 0) break;
esl_sq_Reuse(sq);
}
if (status == eslEOF) esl_fatal("Failed to find sequence %s in file %s\n", key, sqfp->filename);
}
if (do_revcomp == FALSE && newname == NULL && ! esl_sqio_IsAlignment(sqfp->format))
{ /* If we're not manipulating the sequence in any way, and it's not from an alignment file, we can Echo() it. */
if (esl_sqio_Echo(sqfp, sq, ofp) != eslOK) esl_fatal("Echo failed: %s\n", esl_sqfile_GetErrorBuf(sqfp));
}
else
{ /* Otherwise we Write() the parsed version. */
if (do_revcomp && esl_sq_ReverseComplement(sq) != eslOK) esl_fatal("Failed to reverse complement %s; is it a protein?\n", sq->name);
if (newname != NULL) esl_sq_SetName(sq, newname);
esl_sqio_Write(ofp, sq, eslSQFILE_FASTA, FALSE);
}
esl_sq_Destroy(sq);
}
int esl_trans_s2p(ESL_SQ *in, ESL_SQ **out, int frameshift, int rcFlag)
{
// The encoding for this is taken from squid: A=0, C=1, G=2, U/T=3,
// code[0] corresponds to AAA, code[1] is AAC... code[4] is ACA...
// and so on up to 63 being UUU. 64 is a sentinel. Regular 20 amino codes and '*' for stop
// the nucleotide indices match well with the easel alphabet index
// but the actual translation still needs to be hard coded
char code[] = {'K','N','K','N','T','T','T','T','R','S','R','S',
'I','I','M','I','Q','H','Q','H','P','P','P','P',
'R','R','R','R','L','L','L','L','E','D','E','D',
'A','A','A','A','G','G','G','G','V','V','V','V',
'*','Y','*','Y','L','F','L','F','*','C','W','C',
'L','F','L','F'};
int status;
int codon; //progress in counting current codon
char *aaseq; //hold the protein sequence to be output
char *aaptr; //pointer records progress in writing to output
char *readseq; //pointer records progress in reading nucleotide sequence
int read_dg; //index into digital sequence
ESL_ALPHABET *abc = esl_alphabet_Create(eslDNA);
char errbuf[256]; //validateseq demands this
char namestring[256];
(*out) = NULL;
if(frameshift >= in->n) return eslFAIL;
if(!abc) goto ERROR;
//make sure we have a nucleotide sequence; could use esl_abc_ValidateSeq but that wants too
//much boilerplate for the simple bit I need done. doesn't help that i don't care if there are U or T
//characters but that would test against two alphabets
if(in->seq)
{
if(eslOK != esl_abc_ValidateSeq(abc, in->seq, in->n, errbuf)) goto ERROR;
}
else if(in->dsq)
{
if(in->abc->type != eslRNA && in->abc->type != eslDNA) goto ERROR;
}
else
{
goto ERROR;
}
//apply the reverse compliment
if(rcFlag) {if(esl_sq_ReverseComplement(in) != eslOK) goto ERROR;}
ESL_ALLOC(aaseq, (in->n+1) * sizeof(char));
aaptr = aaseq;
if(in->seq) //text sequence
{
//get an alphabet to do the lookup with.
//an ordinary text sequence doesn't have in->abc
//if it has one that is not a standard dna/rna alphabet
//then this code won't work. I wanted to use an alphabet if available, could save some allocating time that way
//if we're calling this repeatedly
//but the compiler complains about "pointer qualifiers" so nevermind
readseq = in->seq+frameshift;
//as long as there are at least 3 nucleotides left, pull and translate another codon
for (; *readseq != '\0' && *(readseq+1) != '\0' && *(readseq+2) != '\0'; readseq += 3)
{
codon = abc->inmap[(int)*(readseq)] * 16 + abc->inmap[(int)*(readseq+1)] * 4 + abc->inmap[(int)*(readseq+2)];
if(codon > 63 || codon < 0) break;
*aaptr = code[codon];
aaptr += 1;
}
*aaptr = '\0';
}
else if(in->dsq) //do it digitally
{
if(in->dsq == NULL) goto ERROR;
read_dg = 1+frameshift; //add one here because digital index 0 is a sentinel
for(;in->dsq[read_dg] != 255 && in->dsq[read_dg+1] != 255 && in->dsq[read_dg+2] != 255; read_dg += 3)
{
codon = in->dsq[read_dg] * 16 + in->dsq[read_dg+1] * 4 + in->dsq[read_dg+2];
if(codon > 63 || codon < 0) break;
*aaptr = code[codon];
aaptr += 1;
}
*aaptr = '\0';
}
else
{
goto ERROR;
}
//modify name to record any reading frame adjustments
sprintf(namestring, "%s_s%d", in->name, frameshift);
if(rcFlag) strcat(namestring, "_rc");
*out = esl_sq_CreateFrom(namestring, aaseq, in->desc, in->acc, in->ss);
if(aaseq != NULL) free(aaseq);
//return the input to its original state
if(rcFlag) {if(esl_sq_ReverseComplement(in) != eslOK) goto ERROR;}
if(abc) esl_alphabet_Destroy(abc);
if(*out) return eslOK;
ERROR:
if(abc) esl_alphabet_Destroy(abc);
if(aaseq != NULL) free(aaseq);
(*out) = NULL;
return eslEMEM;
}
static void
pipeline_thread(void *arg)
{
int i, j;
int status;
int workeridx;
WORKER_INFO *info;
ESL_THREADS *obj;
P7_OM_BLOCK *block;
void *newBlock;
P7_OPROFILE *om = NULL;
P7_SCOREDATA *scoredata = NULL; /* hmm-specific data used by nhmmer */
P7_DOMAIN *dcl;
int seq_len = 0;
int prev_hit_cnt = 0;
#ifdef eslAUGMENT_ALPHABET
ESL_SQ *sq_revcmp = NULL;
#endif /*eslAUGMENT_ALPHABET*/
impl_ThreadInit();
obj = (ESL_THREADS *) arg;
esl_threads_Started(obj, &workeridx);
info = (WORKER_INFO *) esl_threads_GetData(obj, workeridx);
status = esl_workqueue_WorkerUpdate(info->queue, NULL, &newBlock);
if (status != eslOK) esl_fatal("Work queue worker failed");
#ifdef eslAUGMENT_ALPHABET
//reverse complement
if (info->pli->strand != p7_STRAND_TOPONLY && info->qsq->abc->complement != NULL ) {
sq_revcmp = esl_sq_CreateDigital(info->qsq->abc);
esl_sq_Copy(info->qsq,sq_revcmp);
esl_sq_ReverseComplement(sq_revcmp);
info->pli->nres += info->qsq->n;
}
#endif /*eslAUGMENT_ALPHABET*/
/* loop until all blocks have been processed */
block = (P7_OM_BLOCK *) newBlock;
while (block->count > 0)
{
/* Main loop: */
for (i = 0; i < block->count; ++i)
{
om = block->list[i];
seq_len = 0;
p7_pli_NewModel(info->pli, om, info->bg);
p7_bg_SetLength(info->bg, info->qsq->n);
p7_oprofile_ReconfigLength(om, info->qsq->n);
scoredata = p7_hmm_ScoreDataCreate(om, FALSE);
#ifdef eslAUGMENT_ALPHABET
//reverse complement
if (info->pli->strand != p7_STRAND_TOPONLY && info->qsq->abc->complement != NULL )
{
p7_Pipeline_LongTarget(info->pli, om, scoredata, info->bg, sq_revcmp, info->th, 0);
p7_pipeline_Reuse(info->pli); // prepare for next search
seq_len = info->qsq->n;
for (j = prev_hit_cnt; j < info->th->N ; j++)
{
dcl = info->th->unsrt[j].dcl;
// modify hit positions to account for the position of the window in the full sequence
dcl->ienv = seq_len - dcl->ienv + 1;
dcl->jenv = seq_len - dcl->jenv + 1;
dcl->iali = seq_len - dcl->iali + 1;
dcl->jali = seq_len - dcl->jali + 1;
dcl->ad->sqfrom = seq_len - dcl->ad->sqfrom + 1;
dcl->ad->sqto = seq_len - dcl->ad->sqto + 1;
}
}
#endif
if (info->pli->strand != p7_STRAND_BOTTOMONLY) {
p7_Pipeline_LongTarget(info->pli, om, scoredata, info->bg, info->qsq, info->th, 0);
p7_pipeline_Reuse(info->pli);
seq_len += info->qsq->n;
}
for (j = prev_hit_cnt; j < info->th->N ; j++)
{
info->th->unsrt[j].lnP += log((float)seq_len / (float)om->max_length);
info->th->unsrt[j].dcl[0].lnP = info->th->unsrt[j].lnP;
info->th->unsrt[j].sortkey = -1.0 * info->th->unsrt[j].lnP;
info->th->unsrt[j].dcl[0].ad->L = om->M;
}
prev_hit_cnt = info->th->N;
p7_hmm_ScoreDataDestroy(scoredata);
p7_oprofile_Destroy(om);
block->list[i] = NULL;
}
status = esl_workqueue_WorkerUpdate(info->queue, block, &newBlock);
if (status != eslOK) esl_fatal("Work queue worker failed");
block = (P7_OM_BLOCK *) newBlock;
}
#ifdef eslAUGMENT_ALPHABET
esl_sq_Destroy(sq_revcmp);
#endif
status = esl_workqueue_WorkerUpdate(info->queue, block, NULL);
if (status != eslOK) esl_fatal("Work queue worker failed");
esl_threads_Finished(obj, workeridx);
return;
}
static int
serial_loop(WORKER_INFO *info, P7_HMMFILE *hfp)
{
int status;
int i;
int seq_len = 0;
int prev_hit_cnt = 0;
P7_OPROFILE *om = NULL;
P7_SCOREDATA *scoredata = NULL; /* hmm-specific data used by nhmmer */
ESL_ALPHABET *abc = NULL;
P7_DOMAIN *dcl;
#ifdef eslAUGMENT_ALPHABET
ESL_SQ *sq_revcmp = NULL;
if (info->pli->strand != p7_STRAND_TOPONLY && info->qsq->abc->complement != NULL ) {
sq_revcmp = esl_sq_CreateDigital(info->qsq->abc);
esl_sq_Copy(info->qsq,sq_revcmp);
esl_sq_ReverseComplement(sq_revcmp);
info->pli->nres += info->qsq->n;
}
#endif /*eslAUGMENT_ALPHABET*/
/* Main loop: */
while ((status = p7_oprofile_ReadMSV(hfp, &abc, &om)) == eslOK)
{
seq_len = 0;
p7_pli_NewModel(info->pli, om, info->bg);
p7_bg_SetLength(info->bg, info->qsq->n);
p7_oprofile_ReconfigLength(om, info->qsq->n);
scoredata = p7_hmm_ScoreDataCreate(om, FALSE);
#ifdef eslAUGMENT_ALPHABET
//reverse complement
if (info->pli->strand != p7_STRAND_TOPONLY && info->qsq->abc->complement != NULL )
{
p7_Pipeline_LongTarget(info->pli, om, scoredata, info->bg, sq_revcmp, info->th, 0);
p7_pipeline_Reuse(info->pli); // prepare for next search
seq_len = info->qsq->n;
for (i = prev_hit_cnt; i < info->th->N ; i++)
{
dcl = info->th->unsrt[i].dcl;
// modify hit positions to account for the position of the window in the full sequence
dcl->ienv = seq_len - dcl->ienv + 1;
dcl->jenv = seq_len - dcl->jenv + 1;
dcl->iali = seq_len - dcl->iali + 1;
dcl->jali = seq_len - dcl->jali + 1;
dcl->ad->sqfrom = seq_len - dcl->ad->sqfrom + 1;
dcl->ad->sqto = seq_len - dcl->ad->sqto + 1;
}
}
#endif
if (info->pli->strand != p7_STRAND_BOTTOMONLY) {
p7_Pipeline_LongTarget(info->pli, om, scoredata, info->bg, info->qsq, info->th, 0);
p7_pipeline_Reuse(info->pli);
seq_len += info->qsq->n;
}
for (i = prev_hit_cnt; i < info->th->N ; i++)
{
info->th->unsrt[i].lnP += log((float)seq_len / (float)om->max_length);
info->th->unsrt[i].dcl[0].lnP = info->th->unsrt[i].lnP;
info->th->unsrt[i].sortkey = -1.0 * info->th->unsrt[i].lnP;
info->th->unsrt[i].dcl[0].ad->L = om->M;
}
prev_hit_cnt = info->th->N;
p7_oprofile_Destroy(om);
p7_hmm_ScoreDataDestroy(scoredata);
}
esl_alphabet_Destroy(abc);
#ifdef eslAUGMENT_ALPHABET
esl_sq_Destroy(sq_revcmp);
#endif
return status;
}
/* 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 seqs by key
* as we see each line;
* else, without an SSI index, store the keys in a hash, then
* read the entire seq file in a single pass, outputting seqs
* that are in our keylist.
*
* Note that with an SSI index, you get the seqs in the order they
* appear in the <keyfile>, but without an SSI index, you get seqs in
* the order they occur in the seq file.
*/
static void
multifetch(ESL_GETOPTS *go, FILE *ofp, char *keyfile, ESL_SQFILE *sqfp)
{
ESL_KEYHASH *keys = esl_keyhash_Create();
ESL_FILEPARSER *efp = NULL;
int nseq = 0;
int nkeys = 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 seq name on line %d of file %s\n", efp->linenumber, keyfile);
status = esl_keyhash_Store(keys, key, keylen, &keyidx);
if (status == eslEDUP) esl_fatal("seq key %s occurs more than once in file %s\n", key, keyfile);
/* if we have an SSI index, just fetch them as we go. */
if (sqfp->data.ascii.ssi != NULL) { onefetch(go, ofp, key, sqfp); nseq++; }
nkeys++;
}
/* If we don't have an SSI index, we haven't fetched anything yet; do it now. */
if (sqfp->data.ascii.ssi == NULL)
{
ESL_SQ *sq = esl_sq_Create();
while ((status = esl_sqio_Read(sqfp, sq)) == eslOK)
{
if ( (sq->name[0] != '\0' && esl_keyhash_Lookup(keys, sq->name, -1, NULL) == eslOK) ||
(sq->acc[0] != '\0' && esl_keyhash_Lookup(keys, sq->acc, -1, NULL) == eslOK))
{
if (esl_opt_GetBoolean(go, "-r") )
if (esl_sq_ReverseComplement(sq) != eslOK)
esl_fatal("Failed to reverse complement %s\n", sq->name);
esl_sqio_Write(ofp, sq, eslSQFILE_FASTA, FALSE);
nseq++;
}
esl_sq_Reuse(sq);
}
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_sq_Destroy(sq);
}
if (nkeys != nseq) esl_fatal("Tried to retrieve %d keys, but only retrieved %d sequences\n", nkeys, nseq);
if (ofp != stdout) printf("\nRetrieved %d sequences.\n", nseq);
esl_keyhash_Destroy(keys);
esl_fileparser_Close(efp);
return;
}
