/* Open the source sequence database for negative subseqs; * upon return, cfg->dbfp is open (digital, SSI indexed); * cfg->db_maxL and cfg->db_nseq are set. */ static int process_dbfile(struct cfg_s *cfg, char *dbfile, int dbfmt) { ESL_SQ *sq = esl_sq_CreateDigital(cfg->abc); int status; /* Open the sequence file in digital mode */ status = esl_sqfile_OpenDigital(cfg->abc, dbfile, dbfmt, NULL, &(cfg->dbfp)); if (status == eslENOTFOUND) esl_fatal("No such file %s", dbfile); else if (status == eslEFORMAT) esl_fatal("Format of seqfile %s unrecognized.", dbfile); else if (status == eslEINVAL) esl_fatal("Can't autodetect stdin or .gz."); else if (status != eslOK) esl_fatal("Open failed, code %d.", status); /* Read info on each sequence */ cfg->db_nseq = 0; cfg->db_maxL = 0; while ((status = esl_sqio_ReadInfo(cfg->dbfp, sq)) == eslOK) { cfg->db_maxL = ESL_MAX(sq->L, cfg->db_maxL); cfg->db_nseq++; esl_sq_Reuse(sq); } if (status != eslEOF) esl_fatal("Something went wrong with reading the seq db"); /* Open SSI index */ if (esl_sqfile_OpenSSI(cfg->dbfp, NULL) != eslOK) esl_fatal("Failed to open SSI index file"); if (cfg->dbfp->data.ascii.ssi->nprimary != cfg->db_nseq) esl_fatal("oops, nprimary != nseq"); esl_sq_Destroy(sq); return eslOK; }
int main(int argc, char **argv) { ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 1, argc, argv, banner, usage); ESL_ALPHABET *abc = esl_alphabet_Create(eslAMINO); ESL_SQ *sq = esl_sq_CreateDigital(abc); ESL_SQFILE *sqfp = NULL; ESL_HMM *hmm = create_test_hmm(abc); ESL_HMM *bg = create_null_hmm(abc); ESL_HMX *hmx = esl_hmx_Create(400, hmm->M); int format = eslSQFILE_UNKNOWN; char *seqfile = esl_opt_GetArg(go, 1); float fwdsc, nullsc; int status; status = esl_sqfile_OpenDigital(abc, seqfile, format, NULL, &sqfp); if (status == eslENOTFOUND) esl_fatal("No such file."); else if (status == eslEFORMAT) esl_fatal("Format unrecognized."); else if (status != eslOK) esl_fatal("Open failed, code %d.", status); while ((status = esl_sqio_Read(sqfp, sq)) == eslOK) { esl_hmx_GrowTo(hmx, sq->n, hmm->M); esl_hmm_Forward(sq->dsq, sq->n, hmm, hmx, &fwdsc); esl_hmm_Forward(sq->dsq, sq->n, bg, hmx, &nullsc); printf("%-16s %5d %11.4f %8.4f %11.4f %8.4f %11.4f %8.4f\n", sq->name, (int) sq->n, fwdsc * eslCONST_LOG2R, (fwdsc * eslCONST_LOG2R) / sq->n, nullsc * eslCONST_LOG2R, (nullsc * eslCONST_LOG2R) / sq->n, (fwdsc - nullsc) * eslCONST_LOG2R, (fwdsc-nullsc) * eslCONST_LOG2R / sq->n); 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_sqfile_Close(sqfp); esl_sq_Destroy(sq); esl_hmm_Destroy(hmm); esl_hmm_Destroy(bg); esl_hmx_Destroy(hmx); esl_alphabet_Destroy(abc); esl_getopts_Destroy(go); return 0; }
int main(int argc, char **argv) { ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 2, argc, argv, banner, usage); char *hmmfile = esl_opt_GetArg(go, 1); char *seqfile = esl_opt_GetArg(go, 2); ESL_STOPWATCH *w = esl_stopwatch_Create(); ESL_RANDOMNESS*r = esl_randomness_Create(esl_opt_GetInteger(go, "-s")); ESL_ALPHABET*abc = NULL; P7_HMMFILE *hfp = NULL; P7_HMM *hmm = NULL; P7_BG *bg = NULL; P7_PROFILE *gm1, *gm2; int L = 2000; // esl_opt_GetInteger(go, "-L"); int N = esl_opt_GetInteger(go, "-N"); int MaxPart = esl_opt_GetInteger(go, "-M"); __m128 resdata[10]; float *sc1 = (float*) (resdata+0); // uses 1 __m128s ESL_DSQ *dsq = NULL; int i, j; ESL_SQFILE *sqfp = NULL; DATA_STREAM *dstream; struct timeb tbstart, tbend; int sumlengths = 0; if (p7_hmmfile_Open(hmmfile, NULL, &hfp)!= eslOK) p7_Fail("Failed to open HMM file %s", hmmfile); if (p7_hmmfile_Read(hfp, &abc, &hmm) != eslOK) p7_Fail("Failed to read HMM"); bg = p7_bg_Create(abc); p7_bg_SetLength(bg, L); gm1 = p7_profile_Create(hmm->M, abc); gm2 = p7_profile_Create(hmm->M, abc); p7_ProfileConfig(hmm, bg, gm1, L, p7_UNILOCAL); p7_ProfileConfig(hmm, bg, gm2, L, p7_UNILOCAL); dstream = p7_ViterbiStream_Create(gm1); p7_ViterbiStream_Setup(dstream, L+100, MaxPart); // use max L dstream->L = L; // No. of partitions computed without full parallelism ( == no. of threads active while some are idle) int Niters_part = dstream->Npartitions % NTHREADS; // No. of Model lines that could be computed but are wasted by idle threads waiting on the end int Nwasted_threads = dstream->partition * ((NTHREADS-Niters_part) % NTHREADS); // No. of lines in the last partition that go beyond M. It's wasted comp time by a single thread int Nwasted_leftover= (dstream->partition - gm1->M % dstream->partition) % dstream->partition; // Total number of wasted lines int wastedcomp = Nwasted_threads + Nwasted_leftover; // Total number of lines computed and waited for int totalcomp = wastedcomp + gm1->M; // same as: roundtop(gm1->M, dstream->partition * NTHREADS); printf("Viterbi Stream Word with %d Threads, model %s: Modelsize %d, #Segms: %d, SeqL: %d, Nseqs %d, Part %d, #Parts %d\n", NTHREADS, hmmfile, gm1->M, (int) ceil(gm1->M/8.0), L, 8*N, dstream->partition, dstream->Npartitions); printf("Total Comp Lines: %d | Wasted Comp Lines: %d\n", totalcomp, wastedcomp); // for ViterbiFilter P7_OPROFILE *om = p7_oprofile_Create(hmm->M, gm1->abc); p7_oprofile_Convert(gm1, om); P7_OMX *ox = p7_omx_Create(hmm->M, 0, 0); dsq_cmp_t **seqsdb= calloc(8*N+64, sizeof(dsq_cmp_t*)); if(0) { ESL_SQ* sq = esl_sq_CreateDigital(abc); if (esl_sqfile_OpenDigital(abc, seqfile, eslSQFILE_FASTA, NULL, &sqfp) != eslOK) { p7_Fail("Failed to open sequence file\n"); return -1; } for (j = 0; j < 8*N; j++) { int res = esl_sqio_Read(sqfp, sq); if (res != eslOK) { printf("ATENCAO: faltam sequencias\n"); break; } int len = sq->n; dsq = sq->dsq; seqsdb[j] = malloc(sizeof(dsq_cmp_t)); seqsdb[j]->length = len; seqsdb[j]->seq = malloc((len+4)*sizeof(ESL_DSQ)); memcpy(seqsdb[j]->seq, dsq, len+2); sumlengths += len; esl_sq_Reuse(sq); } ftime(&tbstart); N = j/8; printf("N = %d\n", N); // Sort sequences by length qsort(seqsdb, N*8, sizeof(dsq_cmp_t*), compare_seqs); } else if(0) for (i = 0; i < N; i++) { for (j = 0; j < 8; j++) { int len = L - rand()%1000; seqsdb[i*8+j] = malloc(sizeof(dsq_cmp_t)); seqsdb[i*8+j]->seq = malloc(len+4); seqsdb[i*8+j]->length = len; esl_rsq_xfIID(r, bg->f, abc->K, len, seqsdb[i*8+j]->seq); sumlengths += len; } } // double sumerrors = 0; float* results = (float*) alloc_m128_aligned64(N*2+2); ftime(&tbstart); for (j = 0; j < N; j++) for (i = 0; i < N; i++) { // if (i % 10000 == 0) printf("START %d\n", i); p7_ViterbiStream(dstream, seqsdb+i*8, sc1); // memcpy(results+i*8, sc1, 32); } ftime(&tbend); double secs = TIMEDIFF(tbstart,tbend); // printf("Qsort time: %6.3f | Viterbi time: %6.3f\n", TIMEDIFF(tbqsort,tbstart), secs); w->elapsed = w->user = secs; esl_stopwatch_Display(stdout, w, "# Opt CPU time: "); printf("# %.0fM cells in %.1f Mc/s\n", (sumlengths * (double) gm1->M) / 1e6, (sumlengths * (double) gm1->M * 1e-6) / secs); if(0) // compare results against base version for (i = 0; i < 1000 && i < N; i++) { int maxll = 0; float sc2; for (j = 0; j < 8; j++) if (maxll < seqsdb[i*8+j]->length) maxll = seqsdb[i*8+j]->length; // for (j = 0; j < 8; j++) printf("%d ", seqsdb[i*8+j]->length); printf("\n"); // if (i % 10 == 0) printf("i %d\n", i); p7_oprofile_ReconfigRestLength(om, maxll); p7_ReconfigLength(gm2, maxll); // fazer Reconfig aqui para emular compl o VitStream for (j = 0; j < 8; j++) { // p7_ReconfigLength(gm2, seqsdb[i*8+j]->length); // p7_Viterbi_unilocal(seqsdb[i*8+j]->seq, seqsdb[i*8+j]->length, gm2, &sc3); // p7_Viterbi_unilocal_word(seqsdb[i*8+j]->seq, seqsdb[i*8+j]->length, gm2, &sc2); // p7_oprofile_ReconfigLength(om, seqsdb[i*8+j]->length); p7_ViterbiFilter(seqsdb[i*8+j]->seq, seqsdb[i*8+j]->length, om, ox, &sc2); //sumerrors += fabs(sc1[j]- sc2); if (fabs(results[i*8+j] - sc2) > 0.00001) { printf("%3d-%d L %4d: VS %f d %f\t| Base SerI %f\n", i, j, seqsdb[i*8+j]->length, results[i*8+j], fabs(results[i*8+j] - sc2), sc2); getc(stdin); } } } return 0; }
int main(int argc, char **argv) { ESL_GETOPTS *go = p7_CreateDefaultApp(options, 2, argc, argv, banner, usage); char *hmmfile = esl_opt_GetArg(go, 1); char *seqfile = esl_opt_GetArg(go, 2); ESL_ALPHABET *abc = NULL; P7_HMMFILE *hfp = NULL; P7_HMM *hmm = NULL; P7_BG *bg = NULL; P7_PROFILE *gm = NULL; P7_GMX *gx1 = NULL; P7_GMX *gx2 = NULL; ESL_SQ *sq = NULL; ESL_SQFILE *sqfp = NULL; P7_TRACE *tr = NULL; int format = eslSQFILE_UNKNOWN; char errbuf[eslERRBUFSIZE]; float fsc, bsc, vsc; float accscore; int status; /* Read in one HMM */ if (p7_hmmfile_OpenE(hmmfile, NULL, &hfp, NULL) != eslOK) p7_Fail("Failed to open HMM file %s", hmmfile); if (p7_hmmfile_Read(hfp, &abc, &hmm) != eslOK) p7_Fail("Failed to read HMM"); p7_hmmfile_Close(hfp); /* Read in one sequence */ sq = esl_sq_CreateDigital(abc); status = esl_sqfile_OpenDigital(abc, seqfile, format, NULL, &sqfp); if (status == eslENOTFOUND) p7_Fail("No such file."); else if (status == eslEFORMAT) p7_Fail("Format unrecognized."); else if (status == eslEINVAL) p7_Fail("Can't autodetect stdin or .gz."); else if (status != eslOK) p7_Fail("Open failed, code %d.", status); if (esl_sqio_Read(sqfp, sq) != eslOK) p7_Fail("Failed to read sequence"); esl_sqfile_Close(sqfp); /* Configure a profile from the HMM */ bg = p7_bg_Create(abc); p7_bg_SetLength(bg, sq->n); gm = p7_profile_Create(hmm->M, abc); p7_ProfileConfig(hmm, bg, gm, sq->n, p7_LOCAL); /* multihit local: H3 default */ /* Allocations */ gx1 = p7_gmx_Create(gm->M, sq->n); gx2 = p7_gmx_Create(gm->M, sq->n); tr = p7_trace_CreateWithPP(); p7_FLogsumInit(); /* Run Forward, Backward; do OA fill and trace */ p7_GForward (sq->dsq, sq->n, gm, gx1, &fsc); p7_GBackward(sq->dsq, sq->n, gm, gx2, &bsc); p7_GDecoding(gm, gx1, gx2, gx2); /* <gx2> is now the posterior decoding matrix */ p7_GOptimalAccuracy(gm, gx2, gx1, &accscore); /* <gx1> is now the OA matrix */ p7_GOATrace(gm, gx2, gx1, tr); if (esl_opt_GetBoolean(go, "-d")) p7_gmx_Dump(stdout, gx2, p7_DEFAULT); if (esl_opt_GetBoolean(go, "-m")) p7_gmx_Dump(stdout, gx1, p7_DEFAULT); p7_trace_Dump(stdout, tr, gm, sq->dsq); if (p7_trace_Validate(tr, abc, sq->dsq, errbuf) != eslOK) p7_Die("trace fails validation:\n%s\n", errbuf); printf("fwd = %.4f nats\n", fsc); printf("bck = %.4f nats\n", bsc); printf("acc = %.4f (%.2f%%)\n", accscore, accscore * 100. / (float) sq->n); p7_trace_Reuse(tr); p7_GViterbi(sq->dsq, sq->n, gm, gx1, &vsc); p7_GTrace (sq->dsq, sq->n, gm, gx1, tr); p7_trace_SetPP(tr, gx2); p7_trace_Dump(stdout, tr, gm, sq->dsq); printf("vit = %.4f nats\n", vsc); printf("acc = %.4f\n", p7_trace_GetExpectedAccuracy(tr)); /* Cleanup */ esl_sq_Destroy(sq); p7_trace_Destroy(tr); p7_gmx_Destroy(gx1); p7_gmx_Destroy(gx2); p7_profile_Destroy(gm); p7_bg_Destroy(bg); p7_hmm_Destroy(hmm); esl_alphabet_Destroy(abc); esl_getopts_Destroy(go); return 0; }
int main(int argc, char **argv) { ESL_GETOPTS *go = NULL; ESL_ALPHABET *nt_abc = esl_alphabet_Create(eslDNA); ESL_ALPHABET *aa_abc = esl_alphabet_Create(eslAMINO); ESL_GENCODE *gcode = NULL; ESL_GENCODE_WORKSTATE *wrk = NULL; char *seqfile = NULL; int informat = eslSQFILE_UNKNOWN; ESL_SQFILE *sqfp = NULL; int status; /***************************************************************** * command line parsing *****************************************************************/ go = esl_getopts_Create(options); if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK || esl_opt_VerifyConfig(go) != eslOK) { printf("Failed to parse command line: %s\n", go->errbuf); esl_usage(stdout, argv[0], usage); printf("\nTo see more help on available options, do %s -h\n\n", argv[0]); exit(1); } if (esl_opt_GetBoolean(go, "-h") ) { esl_banner(stdout, argv[0], banner); esl_usage (stdout, argv[0], usage); puts("\n where options are:"); esl_opt_DisplayHelp(stdout, go, /*docgroup=*/0, /*indent=*/2, /*textwidth=*/80); puts("\nAvailable NCBI genetic code tables (for -c <id>):"); esl_gencode_DumpAltCodeTable(stdout); exit(0); } if (esl_opt_ArgNumber(go) != 1) { printf("Incorrect number of command line arguments.\n"); esl_usage(stdout, argv[0], usage); printf("\nTo see more help on available options, do %s -h\n\n", argv[0]); exit(1); } seqfile = esl_opt_GetArg(go, 1); if (esl_opt_IsOn(go, "--informat") && (informat = esl_sqio_EncodeFormat(esl_opt_GetString(go, "--informat"))) == eslMSAFILE_UNKNOWN) esl_fatal("%s is not a valid input sequence file format for --informat", esl_opt_GetString(go, "--informat")); status = esl_sqfile_OpenDigital(nt_abc, seqfile, informat, /*env=*/NULL, &sqfp); if (status == eslENOTFOUND) esl_fatal("Failed to find (or open) sequence file %s", seqfile); else if (status == eslEFORMAT) esl_fatal("Failed to recognize format of sequence file %s", seqfile); else if (status != eslOK) esl_fatal("Failure in opening sequence file %s; code %d", seqfile, status); /* A limitation. The esl_sqio_ReadWindow() interface needs to use SSI positioning * to read reverse complement, and that doesn't work on nonrewindable streams. */ if ( esl_opt_GetBoolean(go, "-W") && ! esl_sqfile_IsRewindable(sqfp) && ! esl_opt_GetBoolean(go, "--watson")) esl_fatal("esl-translate can't read reverse complement from a nonrewindable stream (stdin pipe, .gz file, etc)."); /* Set up the genetic code. Default = NCBI 1, the standard code; allow ORFs to start at any aa */ gcode = esl_gencode_Create(nt_abc, aa_abc); esl_gencode_Set(gcode, esl_opt_GetInteger(go, "-c")); // default = 1, the standard genetic code if (esl_opt_GetBoolean(go, "-m")) esl_gencode_SetInitiatorOnlyAUG(gcode); else if (! esl_opt_GetBoolean(go, "-M")) esl_gencode_SetInitiatorAny(gcode); // note this is the default, if neither -m or -M are set /* Set up the workstate structure, which contains both stateful * info about our position in <sqfp> and the DNA <sq>, as well as * one-time config info from options */ wrk = esl_gencode_WorkstateCreate(go, gcode); /* The two styles of main processing loop: */ if (esl_opt_GetBoolean(go, "-W")) do_by_windows(gcode, wrk, sqfp); else do_by_sequences(gcode, wrk, sqfp); esl_gencode_WorkstateDestroy(wrk); esl_sqfile_Close(sqfp); esl_gencode_Destroy(gcode); esl_alphabet_Destroy(aa_abc); esl_alphabet_Destroy(nt_abc); esl_getopts_Destroy(go); return 0; }
int main(int argc, char **argv) { ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 2, argc, argv, banner, usage); char *hmmfile = esl_opt_GetArg(go, 1); char *seqfile = esl_opt_GetArg(go, 2); ESL_STOPWATCH *w = esl_stopwatch_Create(); ESL_RANDOMNESS*r = esl_randomness_Create(esl_opt_GetInteger(go, "-s")); ESL_ALPHABET*abc = NULL; P7_HMMFILE *hfp = NULL; P7_HMM *hmm = NULL; P7_BG *bg = NULL; P7_PROFILE *gm1, *gm2; int L = esl_opt_GetInteger(go, "-L"); int N = esl_opt_GetInteger(go, "-N") / SSE16_NVALS; int MaxPart = esl_opt_GetInteger(go, "-M"); int NROUNDS = esl_opt_GetInteger(go, "-R"); int check = esl_opt_GetBoolean(go, "-c"); __m128 resdata[10]; int i, j; float *sc1 = (float*) resdata; ESL_SQFILE *sqfp = NULL; DATA_COPS16 *dcops; struct timeb tbstart, tbend; int sumlengths = 0; float* results = NULL; srand(time(NULL)); if (p7_hmmfile_Open(hmmfile, NULL, &hfp) != eslOK) p7_Fail("Failed to open HMM file %s", hmmfile); if (p7_hmmfile_Read(hfp, &abc, &hmm) != eslOK) p7_Fail("Failed to read HMM"); bg = p7_bg_Create(abc); p7_bg_SetLength(bg, L); gm1 = p7_profile_Create(hmm->M, abc); gm2 = p7_profile_Create(hmm->M, abc); p7_ProfileConfig(hmm, bg, gm1, L, p7_UNILOCAL); p7_ProfileConfig(hmm, bg, gm2, L, p7_UNILOCAL); dcops = p7_ViterbiCOPSw_Create(gm1); p7_ViterbiCOPSW_Setup(dcops, L+100, MaxPart); // use max L dcops->L = L; int dbsize = SSE16_NVALS*N; SEQ **seqsdb= calloc(dbsize+1, sizeof(SEQ*)); int equallength = 1; if (esl_sqfile_OpenDigital(abc, seqfile, eslSQFILE_FASTA, NULL, &sqfp) == eslOK) { // Use Sequence file ESL_SQ* sq = esl_sq_CreateDigital(abc); int maxseqs, len=0; if (esl_opt_IsDefault(go, "-N")) // N not specified in cmdline maxseqs = INT_MAX; // no limit else maxseqs = SSE16_NVALS*N; // use cmdline limit for (j = 0; j < maxseqs && esl_sqio_Read(sqfp, sq) == eslOK; j++) { if (equallength && sq->n != len && j > 0) equallength = 0; len = sq->n; if (j > dbsize) { seqsdb = realloc(seqsdb, 2*(dbsize+1)*sizeof(SEQ*)); dbsize *= 2; } ESL_DSQ* dsq = sq->dsq; seqsdb[j] = malloc(sizeof(SEQ)); seqsdb[j]->length = len; seqsdb[j]->seq = malloc((len+4)*sizeof(ESL_DSQ)); memcpy(seqsdb[j]->seq, dsq, len+2); sumlengths += len; esl_sq_Reuse(sq); } N = j/SSE16_NVALS; } else // Not found database. Generate random sequences for (i = 0; i < N; i++) for (j = 0; j < SSE16_NVALS; j++) { int len = L; // - rand()%1000; seqsdb[i*SSE16_NVALS+j] = malloc(sizeof(SEQ)); seqsdb[i*SSE16_NVALS+j]->seq = malloc(len+4); seqsdb[i*SSE16_NVALS+j]->length = len; esl_rsq_xfIID(r, bg->f, abc->K, len, seqsdb[i*SSE16_NVALS+j]->seq); sumlengths += len; } printf("Viterbi COPS Word with %d threads, model %s. ModelLen: %d, #Segms: %d, SeqL.: %d, #seqs: %d, Partition: %d, #parts: %d\n", NTHREADS, hmmfile, gm1->M, (int) ceil(gm1->M/SSE16_NVALS), L, SSE16_NVALS*N*NROUNDS, dcops->partition, dcops->Npartitions); /* // No. of partitions computed without full parallelism ( == no. of threads active while some are idle) int Niters_part = dcops->Npartitions % NTHREADS; // No. of Model lines that could be computed but are wasted by idle threads waiting on the end int Nwasted_threads = dcops->partition * ((NTHREADS-Niters_part) % NTHREADS); // No. of lines in the last partition that go beyond M. It's wasted comp time by a single thread int Nwasted_leftover= (dcops->partition - gm1->M % dcops->partition) % dcops->partition; // Total number of wasted lines int wastedcomp = Nwasted_threads + Nwasted_leftover; // Total number of lines computed and waited for int totalcomp = wastedcomp + gm1->M; // same as: roundtop(gm1->M, dcops->partition * NTHREADS); printf("Total Comp Lines: %d | Wasted Comp Lines: %d\n", totalcomp, wastedcomp); */ if (check) results = (float*) alloc_m128_aligned64((N+1)*2); ftime(&tbstart); if (!equallength) { // Sort sequences by length qsort(seqsdb, N*SSE16_NVALS, sizeof(SEQ*), compare_seqs); } for (j = 0; j < NROUNDS; j++) for (i = 0; i < N; i++) { // if (i % 1000 == 0) printf("Seq %d\n", i); p7_ViterbiCOPSw_run(dcops, seqsdb+i*SSE16_NVALS, sc1); if (check) memcpy(results+i*SSE16_NVALS, sc1, 32); // 32 bytes indeed! SSE16_NVALS floats } ftime(&tbend); double secs = TIMEDIFF(tbstart,tbend); w->elapsed = w->user = secs; esl_stopwatch_Display(stdout, w, "# Opt CPU time: "); double compmillioncells = NROUNDS * (double) sumlengths * (double) hmm->M * 1e-6; printf("# %.0fM cells in %.1f Mc/s\n", compmillioncells, compmillioncells / secs); if (check) { P7_OPROFILE *om = p7_oprofile_Create(hmm->M, gm1->abc); p7_oprofile_Convert(gm1, om); P7_OMX *ox = p7_omx_Create(hmm->M, 0, 0); printf("Compare results against base version\n"); for (i = 0; i < N; i++) { int maxll = 0; float sc2; for (j = 0; j < SSE16_NVALS; j++) if (maxll < seqsdb[i*SSE16_NVALS+j]->length) maxll = seqsdb[i*SSE16_NVALS+j]->length; p7_oprofile_ReconfigRestLength(om, maxll); // p7_ReconfigLength(gm2, maxll); // emulate the lock-step inter-sequence reconfigs for (j = 0; j < SSE16_NVALS; j++) { // p7_ReconfigLength(gm2, seqsdb[i*SSE16_NVALS+j]->length); // p7_Viterbi_unilocal(seqsdb[i*SSE16_NVALS+j]->seq, seqsdb[i*SSE16_NVALS+j]->length, gm2, &sc3); // p7_Viterbi_unilocal_word(seqsdb[i*SSE16_NVALS+j]->seq, seqsdb[i*SSE16_NVALS+j]->length, gm2, &sc2); // p7_oprofile_ReconfigLength(om, seqsdb[i*SSE16_NVALS+j]->length); p7_ViterbiFilter(seqsdb[i*SSE16_NVALS+j]->seq, seqsdb[i*SSE16_NVALS+j]->length, om, ox, &sc2); sc2 += 1.0; // -2.0nat optimization, Local to Unilocal mode if (fabs(results[i*SSE16_NVALS+j] - sc2) > 0.0001) { printf("Seq %d Len %4d: %f - %f\tdiff: %f\n", i*SSE16_NVALS+j, seqsdb[i*SSE16_NVALS+j]->length, results[i*SSE16_NVALS+j], sc2, fabs(results[i*SSE16_NVALS+j] - sc2)); } } } } return 0; }
/* serial_master() * The serial version of hmmsearch. * For each query HMM in <hmmdb> search the database for hits. * * A master can only return if it's successful. All errors are handled * immediately and fatally with p7_Fail(). */ static int serial_master(ESL_GETOPTS *go, struct cfg_s *cfg) { FILE *ofp = stdout; /* output file for results (default stdout) */ FILE *tblfp = NULL; /* output stream for tabular per-seq (--tblout) */ FILE *dfamtblfp = NULL; /* output stream for tabular Dfam format (--dfamtblout) */ FILE *aliscoresfp = NULL; /* output stream for alignment scores (--aliscoresout) */ // P7_HMM *hmm = NULL; /* one HMM query */ // P7_SCOREDATA *scoredata = NULL; int seqfmt = eslSQFILE_UNKNOWN; /* format of seqfile */ ESL_SQFILE *sqfp = NULL; /* open seqfile */ P7_HMMFILE *hfp = NULL; /* open HMM database file */ ESL_ALPHABET *abc = NULL; /* sequence alphabet */ P7_OPROFILE *om = NULL; /* target profile */ ESL_STOPWATCH *w = NULL; /* timing */ ESL_SQ *qsq = NULL; /* query sequence */ int nquery = 0; int textw; int status = eslOK; int hstatus = eslOK; int sstatus = eslOK; int i; int ncpus = 0; int infocnt = 0; WORKER_INFO *info = NULL; #ifdef HMMER_THREADS P7_OM_BLOCK *block = NULL; ESL_THREADS *threadObj= NULL; ESL_WORK_QUEUE *queue = NULL; #endif char errbuf[eslERRBUFSIZE]; double window_beta = -1.0 ; int window_length = -1; if (esl_opt_IsUsed(go, "--w_beta")) { if ( ( window_beta = esl_opt_GetReal(go, "--w_beta") ) < 0 || window_beta > 1 ) esl_fatal("Invalid window-length beta value\n"); } if (esl_opt_IsUsed(go, "--w_length")) { if (( window_length = esl_opt_GetInteger(go, "--w_length")) < 4 ) esl_fatal("Invalid window length value\n"); } w = esl_stopwatch_Create(); if (esl_opt_GetBoolean(go, "--notextw")) textw = 0; else textw = esl_opt_GetInteger(go, "--textw"); /* If caller declared an input format, decode it */ if (esl_opt_IsOn(go, "--qformat")) { seqfmt = esl_sqio_EncodeFormat(esl_opt_GetString(go, "--qformat")); if (seqfmt == eslSQFILE_UNKNOWN) p7_Fail("%s is not a recognized input sequence file format\n", esl_opt_GetString(go, "--qformat")); } /* validate options if running as a daemon */ // if (esl_opt_IsOn(go, "--daemon")) { /* running as a daemon, the input format must be type daemon */ // if (seqfmt != eslSQFILE_UNKNOWN && seqfmt != eslSQFILE_DAEMON) // esl_fatal("Input format %s not supported. Must be daemon\n", esl_opt_GetString(go, "--qformat")); // seqfmt = eslSQFILE_DAEMON; // if (strcmp(cfg->seqfile, "-") != 0) esl_fatal("Query sequence file must be '-'\n"); // } /* Open the target profile database to get the sequence alphabet */ status = p7_hmmfile_OpenE(cfg->hmmfile, p7_HMMDBENV, &hfp, errbuf); if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", cfg->hmmfile, errbuf); else if (status == eslEFORMAT) p7_Fail("File format problem, trying to open HMM file %s.\n%s\n", cfg->hmmfile, errbuf); else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, cfg->hmmfile, errbuf); if (! hfp->is_pressed) p7_Fail("Failed to open binary auxfiles for %s: use hmmpress first\n", hfp->fname); hstatus = p7_oprofile_ReadMSV(hfp, &abc, &om); if (hstatus == eslEFORMAT) p7_Fail("bad format, binary auxfiles, %s:\n%s", cfg->hmmfile, hfp->errbuf); else if (hstatus == eslEINCOMPAT) p7_Fail("HMM file %s contains different alphabets", cfg->hmmfile); else if (hstatus != eslOK) p7_Fail("Unexpected error in reading HMMs from %s", cfg->hmmfile); p7_oprofile_Destroy(om); p7_hmmfile_Close(hfp); /* Open the query sequence database */ status = esl_sqfile_OpenDigital(abc, cfg->seqfile, seqfmt, NULL, &sqfp); if (status == eslENOTFOUND) p7_Fail("Failed to open sequence file %s for reading\n", cfg->seqfile); else if (status == eslEFORMAT) p7_Fail("Sequence file %s is empty or misformatted\n", cfg->seqfile); else if (status == eslEINVAL) p7_Fail("Can't autodetect format of a stdin or .gz seqfile"); else if (status != eslOK) p7_Fail("Unexpected error %d opening sequence file %s\n", status, cfg->seqfile); if (sqfp->format > 100) // breaking the law! That range is reserved for msa, for aligned formats p7_Fail("%s contains a multiple sequence alignment; expect unaligned sequences, like FASTA\n", cfg->seqfile); qsq = esl_sq_CreateDigital(abc); /* Open the results output files */ if (esl_opt_IsOn(go, "-o")) { if ((ofp = fopen(esl_opt_GetString(go, "-o"), "w")) == NULL) esl_fatal("Failed to open output file %s for writing\n", esl_opt_GetString(go, "-o")); } if (esl_opt_IsOn(go, "--tblout")) { if ((tblfp = fopen(esl_opt_GetString(go, "--tblout"), "w")) == NULL) esl_fatal("Failed to open tabular per-seq output file %s for writing\n", esl_opt_GetString(go, "--tblfp")); } if (esl_opt_IsOn(go, "--dfamtblout")) { if ((dfamtblfp = fopen(esl_opt_GetString(go, "--dfamtblout"),"w")) == NULL) esl_fatal("Failed to open tabular dfam output file %s for writing\n", esl_opt_GetString(go, "--dfamtblout")); } if (esl_opt_IsOn(go, "--aliscoresout")) { if ((aliscoresfp = fopen(esl_opt_GetString(go, "--aliscoresout"),"w")) == NULL) esl_fatal("Failed to open alignment scores output file %s for writing\n", esl_opt_GetString(go, "--aliscoresout")); } output_header(ofp, go, cfg->hmmfile, cfg->seqfile); #ifdef HMMER_THREADS /* initialize thread data */ if (esl_opt_IsOn(go, "--cpu")) ncpus = esl_opt_GetInteger(go, "--cpu"); else esl_threads_CPUCount(&ncpus); if (ncpus > 0) { threadObj = esl_threads_Create(&pipeline_thread); queue = esl_workqueue_Create(ncpus * 2); } #endif infocnt = (ncpus == 0) ? 1 : ncpus; ESL_ALLOC(info, sizeof(*info) * infocnt); for (i = 0; i < infocnt; ++i) { info[i].bg = p7_bg_Create(abc); #ifdef HMMER_THREADS info[i].queue = queue; #endif } #ifdef HMMER_THREADS for (i = 0; i < ncpus * 2; ++i) { block = p7_oprofile_CreateBlock(BLOCK_SIZE); if (block == NULL) esl_fatal("Failed to allocate sequence block"); status = esl_workqueue_Init(queue, block); if (status != eslOK) esl_fatal("Failed to add block to work queue"); } #endif /* Outside loop: over each query sequence in <seqfile>. */ while ((sstatus = esl_sqio_Read(sqfp, qsq)) == eslOK) { if (sstatus == eslEMEM) p7_Fail("Memory allocation error reading sequence file\n", status); if (sstatus == eslEINCONCEIVABLE) p7_Fail("Unexpected error %d reading sequence file\n", status); // if (qsq->L > NHMMER_MAX_RESIDUE_COUNT) p7_Fail("Input sequence %s in file %s exceeds maximum length of %d bases.\n", qsq->name, cfg->seqfile, NHMMER_MAX_RESIDUE_COUNT); nquery++; esl_stopwatch_Start(w); /* Open the target profile database */ status = p7_hmmfile_OpenE(cfg->hmmfile, p7_HMMDBENV, &hfp, NULL); if (status != eslOK) p7_Fail("Unexpected error %d in opening hmm file %s.\n", status, cfg->hmmfile); #ifdef HMMER_THREADS /* if we are threaded, create a lock to prevent multiple readers */ if (ncpus > 0) { status = p7_hmmfile_CreateLock(hfp); if (status != eslOK) p7_Fail("Unexpected error %d creating lock\n", status); } #endif if (fprintf(ofp, "Query: %s [L=%ld]\n", qsq->name, (long) qsq->n) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed"); if (qsq->acc[0] != 0 && fprintf(ofp, "Accession: %s\n", qsq->acc) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed"); if (qsq->desc[0] != 0 && fprintf(ofp, "Description: %s\n", qsq->desc) < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed"); for (i = 0; i < infocnt; ++i) { /* Create processing pipeline and hit list */ info[i].th = p7_tophits_Create(); info[i].pli = p7_pipeline_Create(go, 100, 100, TRUE, p7_SCAN_MODELS); /* M_hint = 100, L_hint = 100 are just dummies for now */ info[i].pli->hfp = hfp; /* for two-stage input, pipeline needs <hfp> */ p7_pli_NewSeq(info[i].pli, qsq); info[i].qsq = qsq; if ( esl_opt_IsUsed(go, "--toponly") ) info[i].pli->strand = p7_STRAND_TOPONLY; else if ( esl_opt_IsUsed(go, "--bottomonly") ) info[i].pli->strand = p7_STRAND_BOTTOMONLY; else info[i].pli->strand = p7_STRAND_BOTH; #ifdef HMMER_THREADS if (ncpus > 0) esl_threads_AddThread(threadObj, &info[i]); #endif } #ifdef HMMER_THREADS if (ncpus > 0) hstatus = thread_loop(threadObj, queue, hfp); else hstatus = serial_loop(info, hfp); #else hstatus = serial_loop(info, hfp); #endif switch(hstatus) { case eslEFORMAT: p7_Fail("bad file format in HMM file %s", cfg->hmmfile); break; case eslEINCOMPAT: p7_Fail("HMM file %s contains different alphabets", cfg->hmmfile); break; case eslEOF: case eslOK: /* do nothing */ break; default: p7_Fail("Unexpected error in reading HMMs from %s", cfg->hmmfile); } /* merge the results of the search results */ for (i = 1; i < infocnt; ++i) { p7_tophits_Merge(info[0].th, info[i].th); p7_pipeline_Merge(info[0].pli, info[i].pli); p7_pipeline_Destroy(info[i].pli); p7_tophits_Destroy(info[i].th); } /* modify e-value to account for number of models */ for (i = 0; i < info->th->N ; i++) { info->th->unsrt[i].lnP += log((float)info->pli->nmodels); info->th->unsrt[i].dcl[0].lnP = info->th->unsrt[i].lnP; info->th->unsrt[i].sortkey = -1.0 * info->th->unsrt[i].lnP; } /* it's possible to have duplicates based on how viterbi ranges can overlap */ p7_tophits_SortByModelnameAndAlipos(info->th); p7_tophits_RemoveDuplicates(info->th, info->pli->use_bit_cutoffs); /* Print results */ p7_tophits_SortBySortkey(info->th); p7_tophits_Threshold(info->th, info->pli); //tally up total number of hits and target coverage info->pli->n_output = info->pli->pos_output = 0; for (i = 0; i < info->th->N; i++) { if ( (info->th->hit[i]->flags & p7_IS_REPORTED) || info->th->hit[i]->flags & p7_IS_INCLUDED) { info->pli->n_output++; info->pli->pos_output += abs(info->th->hit[i]->dcl[0].jali - info->th->hit[i]->dcl[0].iali) + 1; } } p7_tophits_Targets(ofp, info->th, info->pli, textw); if (fprintf(ofp, "\n\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed"); p7_tophits_Domains(ofp, info->th, info->pli, textw); if (fprintf(ofp, "\n\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed"); if (tblfp) p7_tophits_TabularTargets(tblfp, qsq->name, qsq->acc, info->th, info->pli, (nquery == 1)); if (dfamtblfp) p7_tophits_TabularXfam(dfamtblfp, qsq->name, NULL, info->th, info->pli); if (aliscoresfp) p7_tophits_AliScores(aliscoresfp, qsq->name, info->th ); esl_stopwatch_Stop(w); info->pli->nseqs = 1; p7_pli_Statistics(ofp, info->pli, w); if (fprintf(ofp, "//\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed"); fflush(ofp); p7_hmmfile_Close(hfp); p7_pipeline_Destroy(info->pli); p7_tophits_Destroy(info->th); esl_sq_Reuse(qsq); } if (sstatus == eslEFORMAT) esl_fatal("Parse failed (sequence file %s):\n%s\n", sqfp->filename, esl_sqfile_GetErrorBuf(sqfp)); else if (sstatus != eslEOF) esl_fatal("Unexpected error %d reading sequence file %s", sstatus, sqfp->filename); /* Terminate outputs - any last words? */ if (tblfp) p7_tophits_TabularTail(tblfp, "hmmscan", p7_SCAN_MODELS, cfg->seqfile, cfg->hmmfile, go); if (ofp) { if (fprintf(ofp, "[ok]\n") < 0) ESL_EXCEPTION_SYS(eslEWRITE, "write failed"); } /* Cleanup - prepare for successful exit */ for (i = 0; i < infocnt; ++i) p7_bg_Destroy(info[i].bg); #ifdef HMMER_THREADS if (ncpus > 0) { esl_workqueue_Reset(queue); while (esl_workqueue_Remove(queue, (void **) &block) == eslOK) p7_oprofile_DestroyBlock(block); esl_workqueue_Destroy(queue); esl_threads_Destroy(threadObj); } #endif free(info); esl_sq_Destroy(qsq); esl_stopwatch_Destroy(w); esl_alphabet_Destroy(abc); esl_sqfile_Close(sqfp); if (ofp != stdout) fclose(ofp); if (tblfp) fclose(tblfp); if (dfamtblfp) fclose(dfamtblfp); if (aliscoresfp) fclose(aliscoresfp); return eslOK; ERROR: if (ofp != stdout) fclose(ofp); if (tblfp) fclose(tblfp); if (dfamtblfp) fclose(dfamtblfp); if (aliscoresfp) fclose(aliscoresfp); return status; }
int main(int argc, char **argv) { ESL_GETOPTS *go = esl_getopts_CreateDefaultApp(options, 2, argc, argv, banner, usage); char *hmmfile = esl_opt_GetArg(go, 1); char *seqfile = esl_opt_GetArg(go, 2); ESL_ALPHABET *abc = NULL; P7_HMMFILE *hfp = NULL; P7_HMM *hmm = NULL; P7_BG *bg = NULL; ESL_SQFILE *sqfp = NULL; int format = eslSQFILE_UNKNOWN; ESL_SQ *sq = NULL; float nullsc, filtersc, H; int status; /* Read one HMM from <hmmfile> */ if (p7_hmmfile_Open(hmmfile, NULL, &hfp) != eslOK) p7_Fail("Failed to open HMM file %s", hmmfile); if (p7_hmmfile_Read(hfp, &abc, &hmm) != eslOK) p7_Fail("Failed to read HMM"); p7_hmmfile_Close(hfp); /* Open <seqfile> for reading */ status = esl_sqfile_OpenDigital(abc, seqfile, format, NULL, &sqfp); if (status == eslENOTFOUND) esl_fatal("No such file."); else if (status == eslEFORMAT) esl_fatal("Format unrecognized."); else if (status != eslOK) esl_fatal("Open failed, code %d.", status); sq = esl_sq_CreateDigital(abc); bg = p7_bg_Create(abc); p7_bg_SetFilterByHMM(bg, hmm); H = esl_vec_FEntropy(bg->f, bg->abc->K); printf("bg iid H = %.4f\n", H); H = esl_vec_FEntropy(bg->mcomp, bg->abc->K); printf("modelcomp H = %.4f\n", H); while ((status = esl_sqio_Read(sqfp, sq)) == eslOK) { p7_bg_SetLength(bg, sq->n); p7_bg_NullOne (bg, sq->dsq, sq->n, &nullsc); p7_bg_FilterScore(bg, sq->dsq, sq->n, &filtersc); printf("%-20s %5d %8.5f %8.5f %8.5f\n", sq->name, (int) sq->n, nullsc, filtersc, filtersc-nullsc); esl_sq_Reuse(sq); } if (status == eslEFORMAT) esl_fatal("Parse failed (sequence file %s line %" PRId64 "):\n%s\n", sqfp->filename, sqfp->linenumber, sqfp->errbuf); else if (status != eslEOF) esl_fatal("Unexpected error %d reading sequence file %s", status, sqfp->filename); esl_sqfile_Close(sqfp); esl_sq_Destroy(sq); p7_bg_Destroy(bg); p7_hmm_Destroy(hmm); esl_alphabet_Destroy(abc); esl_getopts_Destroy(go); return 0; }
int main(int argc, char **argv) { ESL_GETOPTS *go = NULL; /* application configuration */ char *hmmfile = NULL; /* HMM file name */ char *seqfile = NULL; /* sequence file name */ char *mapfile = NULL; /* optional mapped MSA file name */ int infmt = eslSQFILE_UNKNOWN; int outfmt = eslMSAFILE_STOCKHOLM; P7_HMMFILE *hfp = NULL; /* open HMM file */ ESL_SQFILE *sqfp = NULL; /* open sequence file */ char *outfile = NULL; /* output filename */ FILE *ofp = stdout; /* output stream */ ESL_SQ **sq = NULL; /* array of sequences */ void *p = NULL; /* tmp ptr for reallocation */ int nseq = 0; /* # of sequences in <seqfile> */ int mapseq = 0; /* # of sequences in mapped MSA */ int totseq = 0; /* # of seqs in all sources */ ESL_ALPHABET *abc = NULL; /* alphabet (set from the HMM file)*/ P7_HMM *hmm = NULL; P7_TRACE **tr = NULL; /* array of tracebacks */ ESL_MSA *msa = NULL; /* resulting multiple alignment */ int msaopts = 0; /* flags to p7_tracealign_Seqs() */ int idx; /* counter over seqs, traces */ int status; /* easel/hmmer return code */ char errbuf[eslERRBUFSIZE]; /* Parse the command line */ go = esl_getopts_Create(options); if (esl_opt_ProcessCmdline(go, argc, argv) != eslOK) cmdline_failure(argv[0], "Failed to parse command line: %s\n", go->errbuf); if (esl_opt_VerifyConfig(go) != eslOK) cmdline_failure(argv[0], "Error in configuration: %s\n", go->errbuf); if (esl_opt_GetBoolean(go, "-h") ) cmdline_help (argv[0], go); if (esl_opt_ArgNumber(go) != 2) cmdline_failure(argv[0], "Incorrect number of command line arguments.\n"); hmmfile = esl_opt_GetArg(go, 1); seqfile = esl_opt_GetArg(go, 2); if (strcmp(hmmfile, "-") == 0 && strcmp(seqfile, "-") == 0) cmdline_failure(argv[0], "Either <hmmfile> or <seqfile> may be '-' (to read from stdin), but not both.\n"); msaopts |= p7_ALL_CONSENSUS_COLS; /* default as of 3.1 */ if (esl_opt_GetBoolean(go, "--trim")) msaopts |= p7_TRIM; /* If caller declared an input format, decode it */ if (esl_opt_IsOn(go, "--informat")) { infmt = esl_sqio_EncodeFormat(esl_opt_GetString(go, "--informat")); if (infmt == eslSQFILE_UNKNOWN) cmdline_failure(argv[0], "%s is not a recognized input sequence file format\n", esl_opt_GetString(go, "--informat")); } /* Determine output alignment file format */ outfmt = eslx_msafile_EncodeFormat(esl_opt_GetString(go, "--outformat")); if (outfmt == eslMSAFILE_UNKNOWN) cmdline_failure(argv[0], "%s is not a recognized output MSA file format\n", esl_opt_GetString(go, "--outformat")); /* Open output stream */ if ( (outfile = esl_opt_GetString(go, "-o")) != NULL) { if ((ofp = fopen(outfile, "w")) == NULL) cmdline_failure(argv[0], "failed to open -o output file %s for writing\n", outfile); } /* If caller forced an alphabet on us, create the one the caller wants */ if (esl_opt_GetBoolean(go, "--amino")) abc = esl_alphabet_Create(eslAMINO); else if (esl_opt_GetBoolean(go, "--dna")) abc = esl_alphabet_Create(eslDNA); else if (esl_opt_GetBoolean(go, "--rna")) abc = esl_alphabet_Create(eslRNA); /* Read one HMM, and make sure there's only one. */ status = p7_hmmfile_OpenE(hmmfile, NULL, &hfp, errbuf); if (status == eslENOTFOUND) p7_Fail("File existence/permissions problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf); else if (status == eslEFORMAT) p7_Fail("File format problem in trying to open HMM file %s.\n%s\n", hmmfile, errbuf); else if (status != eslOK) p7_Fail("Unexpected error %d in opening HMM file %s.\n%s\n", status, hmmfile, errbuf); status = p7_hmmfile_Read(hfp, &abc, &hmm); if (status == eslEFORMAT) p7_Fail("Bad file format in HMM file %s:\n%s\n", hfp->fname, hfp->errbuf); else if (status == eslEINCOMPAT) p7_Fail("HMM in %s is not in the expected %s alphabet\n", hfp->fname, esl_abc_DecodeType(abc->type)); else if (status == eslEOF) p7_Fail("Empty HMM file %s? No HMM data found.\n", hfp->fname); else if (status != eslOK) p7_Fail("Unexpected error in reading HMMs from %s\n", hfp->fname); status = p7_hmmfile_Read(hfp, &abc, NULL); if (status != eslEOF) p7_Fail("HMM file %s does not contain just one HMM\n", hfp->fname); p7_hmmfile_Close(hfp); /* We're going to build up two arrays: sequences and traces. * If --mapali option is chosen, the first set of sequences/traces is from the provided alignment */ if ( (mapfile = esl_opt_GetString(go, "--mapali")) != NULL) { map_alignment(mapfile, hmm, &sq, &tr, &mapseq); } totseq = mapseq; /* Read digital sequences into an array (possibly concat'ed onto mapped seqs) */ status = esl_sqfile_OpenDigital(abc, seqfile, infmt, NULL, &sqfp); if (status == eslENOTFOUND) p7_Fail("Failed to open sequence file %s for reading\n", seqfile); else if (status == eslEFORMAT) p7_Fail("Sequence file %s is empty or misformatted\n", seqfile); else if (status != eslOK) p7_Fail("Unexpected error %d opening sequence file %s\n", status, seqfile); ESL_RALLOC(sq, p, sizeof(ESL_SQ *) * (totseq + 1)); sq[totseq] = esl_sq_CreateDigital(abc); nseq = 0; while ((status = esl_sqio_Read(sqfp, sq[totseq+nseq])) == eslOK) { nseq++; ESL_RALLOC(sq, p, sizeof(ESL_SQ *) * (totseq+nseq+1)); sq[totseq+nseq] = esl_sq_CreateDigital(abc); } 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_sqfile_Close(sqfp); totseq += nseq; /* Remaining initializations, including trace array allocation */ ESL_RALLOC(tr, p, sizeof(P7_TRACE *) * totseq); for (idx = mapseq; idx < totseq; idx++) tr[idx] = p7_trace_CreateWithPP(); p7_tracealign_computeTraces(hmm, sq, mapseq, totseq - mapseq, tr); p7_tracealign_Seqs(sq, tr, totseq, hmm->M, msaopts, hmm, &msa); eslx_msafile_Write(ofp, msa, outfmt); for (idx = 0; idx <= totseq; idx++) esl_sq_Destroy(sq[idx]); /* including sq[nseq] because we overallocated */ for (idx = 0; idx < totseq; idx++) p7_trace_Destroy(tr[idx]); free(sq); free(tr); esl_msa_Destroy(msa); p7_hmm_Destroy(hmm); if (ofp != stdout) fclose(ofp); esl_alphabet_Destroy(abc); esl_getopts_Destroy(go); return eslOK; ERROR: return status; }
int main(int argc, char **argv) { ESL_GETOPTS *go = p7_CreateDefaultApp(options, 2, argc, argv, banner, usage); ESL_ALPHABET *abc = esl_alphabet_Create(eslAMINO); char *qfile = esl_opt_GetArg(go, 1); char *tfile = esl_opt_GetArg(go, 2); ESL_SQFILE *qfp = NULL; ESL_SQFILE *tfp = NULL; ESL_SQ *qsq = esl_sq_CreateDigital(abc); ESL_SQ *tsq = esl_sq_CreateDigital(abc); ESL_SCOREMATRIX *S = esl_scorematrix_Create(abc); ESL_DMATRIX *Q = NULL; P7_BG *bg = p7_bg_Create(abc); P7_HMM *hmm = NULL; P7_PROFILE *gm = NULL; P7_REFMX *vit = p7_refmx_Create(200, 400); /* will grow as needed */ double *fa = malloc(sizeof(double) * abc->K); double popen = 0.02; double pextend = 0.4; double lambda; float vsc; float nullsc; int status; esl_composition_BL62(fa); esl_vec_D2F(fa, abc->K, bg->f); esl_scorematrix_Set("BLOSUM62", S); esl_scorematrix_ProbifyGivenBG(S, fa, fa, &lambda, &Q); esl_scorematrix_JointToConditionalOnQuery(abc, Q); if (esl_sqfile_OpenDigital(abc, qfile, eslSQFILE_UNKNOWN, NULL, &qfp) != eslOK) esl_fatal("failed to open %s", qfile); if (esl_sqio_Read(qfp, qsq) != eslOK) esl_fatal("failed to read query seq"); p7_Seqmodel(abc, qsq->dsq, qsq->n, qsq->name, Q, bg->f, popen, pextend, &hmm); p7_hmm_SetComposition(hmm); p7_hmm_SetConsensus(hmm, qsq); gm = p7_profile_Create(hmm->M, abc); p7_profile_ConfigUnilocal(gm, hmm, bg, 400); if (esl_sqfile_OpenDigital(abc, tfile, eslSQFILE_UNKNOWN, NULL, &tfp) != eslOK) esl_fatal("failed to open %s", tfile); while ((status = esl_sqio_Read(tfp, tsq)) == eslOK) { p7_bg_SetLength (bg, tsq->n); p7_profile_SetLength(gm, tsq->n); p7_ReferenceViterbi(tsq->dsq, tsq->n, gm, vit, NULL, &vsc); p7_bg_NullOne(bg, tsq->dsq, tsq->n, &nullsc); printf("%.4f %-25s %-25s\n", (vsc - nullsc) / eslCONST_LOG2, tsq->name, gm->name); esl_sq_Reuse(tsq); p7_refmx_Reuse(vit); } p7_refmx_Destroy(vit); p7_profile_Destroy(gm); p7_hmm_Destroy(hmm); p7_bg_Destroy(bg); esl_dmatrix_Destroy(Q); esl_scorematrix_Destroy(S); free(fa); esl_sq_Destroy(qsq); esl_sq_Destroy(tsq); esl_sqfile_Close(qfp); esl_sqfile_Close(tfp); esl_alphabet_Destroy(abc); esl_getopts_Destroy(go); return 0; }