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);
}
/* Fetch in a random sequence of length <L> from the the pre-digitized
* concatenated sequence database, select a random subseq, shuffle it
* by the chosen algorithm; set dsq[1..L] to the resulting randomized
* segment.
*
* If <logfp> is non-NULL, append one or more "<sqname> <from> <to>"
* fields to current line, to record where the random segment was
* selected from. This is useful in cases where we want to track back
* the origin of a high-scoring segment, in case the randomization
* wasn't good enough to obscure the identity of a segment.
*
*/
static int
set_random_segment(ESL_GETOPTS *go, struct cfg_s *cfg, FILE *logfp, ESL_DSQ *dsq, int L)
{
ESL_SQ *sq = esl_sq_CreateDigital(cfg->abc);
int minDPL = esl_opt_GetInteger(go, "--minDPL");
int db_dependent = (esl_opt_GetBoolean(go, "--iid") == TRUE ? FALSE : TRUE);
char *pkey = NULL;
int start, end;
int64_t Lseq;
int status;
if (L==0) return eslOK;
if (L > cfg->db_maxL) esl_fatal("can't fetch a segment of length %d; database max is %d\n", L, cfg->db_maxL);
/* fetch a random subseq from the source database */
esl_sq_GrowTo(sq, L);
if (db_dependent)
{
do {
if (pkey != NULL) free(pkey);
if (esl_ssi_FindNumber(cfg->dbfp->data.ascii.ssi, esl_rnd_Roll(cfg->r, cfg->db_nseq), NULL, NULL, NULL, &Lseq, &pkey) != eslOK)
esl_fatal("failed to look up a random seq");
} while (Lseq < L);
start = 1 + esl_rnd_Roll(cfg->r, Lseq-L);
end = start + L - 1;
if (esl_sqio_FetchSubseq(cfg->dbfp, pkey, start, end, sq) != eslOK) esl_fatal("failed to fetch subseq");
esl_sq_ConvertDegen2X(sq);
}
/* log sequence source info: <name> <start> <end> */
if (logfp != NULL && db_dependent)
fprintf(logfp, " %-15s %5d %5d", pkey, start, end);
/* Now apply the appropriate randomization algorithm */
if (esl_opt_GetBoolean(go, "--mono")) status = esl_rsq_XShuffle (cfg->r, sq->dsq, L, sq->dsq);
else if (esl_opt_GetBoolean(go, "--di")) {
if (L < minDPL) status = esl_rsq_XShuffle (cfg->r, sq->dsq, L, sq->dsq);
else status = esl_rsq_XShuffleDP(cfg->r, sq->dsq, L, cfg->abc->Kp, sq->dsq);
}
else if (esl_opt_GetBoolean(go, "--markov0")) status = esl_rsq_XMarkov0 (cfg->r, sq->dsq, L, cfg->abc->Kp, sq->dsq);
else if (esl_opt_GetBoolean(go, "--markov1")) status = esl_rsq_XMarkov1 (cfg->r, sq->dsq, L, cfg->abc->Kp, sq->dsq);
else if (esl_opt_GetBoolean(go, "--reverse")) status = esl_rsq_XReverse (sq->dsq, L, sq->dsq);
else if (esl_opt_GetBoolean(go, "--iid")) status = esl_rsq_xIID (cfg->r, cfg->fq, cfg->abc->K, L, sq->dsq);
else status = eslEINCONCEIVABLE;
if (status != eslOK) esl_fatal("esl's shuffling failed");
memcpy(dsq, sq->dsq+1, sizeof(ESL_DSQ) * L);
esl_sq_Destroy(sq);
free(pkey);
return eslOK;
}
