bool updateModel, calcExpectedWeights;
bool genGibbsOut; // generate file for Gibbs sampler
-char refName[STRLEN], imdName[STRLEN], outName[STRLEN];
+char refName[STRLEN], outName[STRLEN];
+char imdName[STRLEN], statName[STRLEN];
char refF[STRLEN], groupF[STRLEN], cntF[STRLEN], tiF[STRLEN];
char mparamsF[STRLEN], bmparamsF[STRLEN];
char modelF[STRLEN], thetaF[STRLEN];
if (!readers[i]->locate(curnr)) { fprintf(stderr, "Read indices files do not match!\n"); exit(-1); }
//assert(readers[i]->locate(curnr));
- while (nrLeft > ntLeft && hitvs[i]->getNHits() < nhT) {
+ while (nrLeft > ntLeft && (i == nThreads - 1 || hitvs[i]->getNHits() < nhT)) {
if (!hitvs[i]->read(fin)) { fprintf(stderr, "Cannot read alignments from .dat file!\n"); exit(-1); }
//assert(hitvs[i]->read(fin));
--nrLeft;
char outF[STRLEN];
FILE *fo;
- sprintf(modelF, "%s.model", outName);
+ sprintf(modelF, "%s.model", statName);
model.write(modelF);
//calculate tau values
delete[] mhps;
}
+int tmp_n;
+
inline bool doesUpdateModel(int ROUND) {
- //return false; // never update, for debugging only
- return ROUND <= 20 || ROUND % 100 == 0;
+ // return ROUND <= 20 || ROUND % 100 == 0;
+ return ROUND <= 10;
}
//Including initialize, algorithm and results saving
}
if (verbose) printf("ROUND = %d, SUM = %.15g, bChange = %f, totNum = %d\n", ROUND, sum, bChange, totNum);
- } while (ROUND < MIN_ROUND || totNum > 0 && ROUND < MAX_ROUND);
+ } while (ROUND < MIN_ROUND || (totNum > 0 && ROUND < MAX_ROUND));
//while (ROUND < MAX_ROUND);
if (totNum > 0) fprintf(stderr, "Warning: RSEM reaches %d iterations before meeting the convergence criteria.\n", MAX_ROUND);
}
}
fclose(fo);
-
- char scoreF[STRLEN];
- sprintf(scoreF, "%s.ns", imdName);
- fo = fopen(scoreF, "w");
- fprintf(fo, "%.15g\n", model.getLogP());
- fclose(fo);
}
- sprintf(thetaF, "%s.theta", outName);
+ sprintf(thetaF, "%s.theta", statName);
fo = fopen(thetaF, "w");
fprintf(fo, "%d\n", M + 1);
//calculate expected weights and counts using learned parameters
updateModel = false; calcExpectedWeights = true;
+ for (int i = 0; i <= M; i++) probv[i] = theta[i];
for (int i = 0; i < nThreads; i++) {
rc = pthread_create(&threads[i], &attr, E_STEP<ReadType, HitType, ModelType>, (void*)(&fparams[i]));
if (rc != 0) { fprintf(stderr, "Cannot create thread %d when calculate expected weights! (numbered from 0)\n", i); exit(-1); }
bool quiet = false;
if (argc < 5) {
- printf("Usage : rsem-run-em refName read_type imdName outName [-p #Threads] [-b samInpType samInpF has_fn_list_? [fn_list]] [-q] [--gibbs-out]\n\n");
+ printf("Usage : rsem-run-em refName read_type sampleName sampleToken [-p #Threads] [-b samInpType samInpF has_fn_list_? [fn_list]] [-q] [--gibbs-out]\n\n");
printf(" refName: reference name\n");
printf(" read_type: 0 single read without quality score; 1 single read with quality score; 2 paired-end read without quality score; 3 paired-end read with quality score.\n");
- printf(" imdName: name for all upstream/downstream user-unseen files. (different files have different suffices)\n");
- printf(" outName: name for all output files. (different files have different suffices)\n");
+ printf(" sampleName: sample's name, including the path\n");
+ printf(" sampleToken: sampleName excludes the path\n");
printf(" -p: number of threads which user wants to use. (default: 1)\n");
printf(" -b: produce bam format output file. (default: off)\n");
printf(" -q: set it quiet\n");
strcpy(refName, argv[1]);
read_type = atoi(argv[2]);
- strcpy(imdName, argv[3]);
- strcpy(outName, argv[4]);
+ strcpy(outName, argv[3]);
+ sprintf(imdName, "%s.temp/%s", argv[3], argv[4]);
+ sprintf(statName, "%s.stat/%s", argv[3], argv[4]);
nThreads = 1;
sprintf(tiF, "%s.ti", refName);
transcripts.readFrom(tiF);
- sprintf(cntF, "%s.cnt", imdName);
+ sprintf(cntF, "%s.cnt", statName);
fin.open(cntF);
if (!fin.is_open()) { fprintf(stderr, "Cannot open %s! It may not exist.\n", cntF); exit(-1); }
fin>>N0>>N1>>N2>>N_tot;