Beispiel #1
0
bam_streamer::
~bam_streamer()
{
    if (NULL != _biter) bam_iter_destroy(_biter);
    if (NULL != _bidx) bam_index_destroy(_bidx);
    if (NULL != _bfp) samclose(_bfp);
}
Beispiel #2
0
static void * downloadBamFile(void * args) {
	BamReaderData * data = (BamReaderData *) args;
	int j, tid, cnt, pos, n_plp;
	const bam_pileup1_t *plp;

	while (bam_mplp_auto(data->iter, &tid, &pos, &n_plp, &plp) > 0) {
		// Count reads in pileup
		cnt = 0;
		const bam_pileup1_t *p = plp;
		for (j = 0; j < n_plp; ++j) {
			//if (bam1_qual(p->b)[p->qpos] >= data->conf->min_baseQ) 
				cnt++;
			p++;
		}

		// Its a wrap:
		char * chrom = data->data->h->target_name[tid];

		if (data->stop > 0 && (strcmp(chrom, data->chrom) == 0 && pos >= data->stop))
			break;

		// +1 to account for 0-based indexing in BAMs:
		if (pushValuesToBuffer(data->bufferedReaderData, chrom, pos+1, pos+2, cnt))
			break;
	}

	bam_iter_destroy(data->data->iter);
	endBufferedSignal(data->bufferedReaderData);
	return NULL;
}
Beispiel #3
0
void SR_BamInStreamFree(SR_BamInStream* pBamInStream)
{
    if (pBamInStream != NULL)
    {
        kh_destroy(queryName, pBamInStream->pNameHashes[PREV_BIN]);
        kh_destroy(queryName, pBamInStream->pNameHashes[CURR_BIN]);

        if (pBamInStream->pRetLists != NULL)
	  free(pBamInStream->pRetLists);
        if (pBamInStream->pAlgnTypes != NULL)
	  free(pBamInStream->pAlgnTypes);
        SR_BamMemPoolFree(pBamInStream->pMemPool);

        bam_close(pBamInStream->fpBamInput);
        bam_index_destroy(pBamInStream->pBamIndex);

	if (pBamInStream->pBamIterator != NULL) {
	  bam_iter_destroy(*(pBamInStream->pBamIterator));
	  free(pBamInStream->pBamIterator);
	  pBamInStream->pBamIterator = NULL;
	}

        free(pBamInStream);
    }
}
Beispiel #4
0
void closeBamFile(BamReaderData * data) {
	// Seriously, does samtools not provide any convience destructors!??
	bam_mplp_destroy(data->iter);
	//bam_header_destroy(data->data->h);
	bam_close(data->data->fp);
	if (data->data->iter) 
		bam_iter_destroy(data->data->iter);
	free(data->data); 
	bam_index_destroy(data->idx);
}
Beispiel #5
0
void hash_reads( table* T, const char* reads_fn, interval_stack* is )
{
    samfile_t* reads_f = samopen( reads_fn, "rb", NULL );
    if( reads_f == NULL ) {
        failf( "Can't open bam file '%s'.", reads_fn );
    }

    bam_index_t* reads_index = bam_index_load( reads_fn );
    if( reads_index == NULL ) {
        failf( "Can't open bam index '%s.bai'.", reads_fn );
    }

    bam_init_header_hash( reads_f->header );

    table_create( T, reads_f->header->n_targets );
    T->seq_names = (char**)malloc( sizeof(char*) * reads_f->header->n_targets );
    size_t k;
    for( k = 0; k < reads_f->header->n_targets; k++ ) {
        T->seq_names[k] = strdup(reads_f->header->target_name[k]);
    }

    log_puts( LOG_MSG, "hashing reads ... \n" );
    log_indent();
    bam_iter_t read_iter;
    bam1_t* read = bam_init1();
    int tid;

    interval_stack::iterator i;
    for( i = is->begin(); i != is->end(); i++ ) {
        tid = bam_get_tid( reads_f->header, i->seqname );
        if( tid < 0 ) continue;

        read_iter = bam_iter_query( reads_index, tid,
                                    i->start, i->end );

        while( bam_iter_read( reads_f->x.bam, read_iter, read ) >= 0 ) {
            if( bam1_strand(read) == i->strand ) {
                table_inc( T, read );
            }
        }

        bam_iter_destroy(read_iter);
    }

    bam_destroy1(read);

    log_unindent();
    log_printf( LOG_MSG, "done. (%zu unique reads hashed)\n", T->m );


    bam_index_destroy(reads_index);
    samclose(reads_f);
}
Beispiel #6
0
int sam_fetch(char *ifn, char *ofn, char *reg, void *data, sam_fetch_f func) {
	int ret = 0;
	samfile_t *in = samopen(ifn, "rb", 0);
	samfile_t *out = 0;
	if (ofn) out = samopen(ofn, "wb", in->header);

	if (reg) {
		bam_index_t *idx = bam_index_load(ifn);
		if (idx == 0) {
			fprintf(stderr, "[%s:%d] Random alignment retrieval only works for indexed BAM files.\n",
							__func__, __LINE__);
			exit(1);
		}
		int tid, beg, end;
		bam_parse_region(in->header, reg, &tid, &beg, &end);
		if (tid < 0) {
			fprintf(stderr, "[%s:%d] Region \"%s\" specifies an unknown reference name. \n",
							__func__, __LINE__, reg);
			exit(1);
		}
		bam_iter_t iter;
		bam1_t *b = bam_init1();
		iter = bam_iter_query(idx, tid, beg, end);
		while ((ret = bam_iter_read(in->x.bam, iter, b)) >= 0) func(b, in, out, data);
		bam_iter_destroy(iter);
		bam_destroy1(b);
		bam_index_destroy(idx);
	} else {
		bam1_t *b = bam_init1();
		while ((ret = samread(in, b)) >= 0) func(b, in, out, data);
		bam_destroy1(b);
	}
	if (out) samclose(out);
	samclose(in);
			
	if (ret != -1) {					/* truncated is -2 */
		fprintf(stderr, "[%s:%d] Alignment retrieval failed due to truncated file\n",
						__func__, __LINE__);
		exit(1);
	}

	return ret;
}
Beispiel #7
0
static void SR_BamInStreamReset(SR_BamInStream* pBamInStream)
{
    pBamInStream->pNewNode = NULL;

    if (pBamInStream->pBamIterator != NULL) {
	bam_iter_destroy(*(pBamInStream->pBamIterator));
	free(pBamInStream->pBamIterator);
	pBamInStream->pBamIterator = NULL;
    }

    pBamInStream->currBinPos = NO_QUERY_YET;
    pBamInStream->currRefID = NO_QUERY_YET;

    kh_clear(queryName, pBamInStream->pNameHashes[PREV_BIN]);
    kh_clear(queryName, pBamInStream->pNameHashes[CURR_BIN]);

    SR_BamListReset(&(pBamInStream->pAlgnLists[PREV_BIN]), pBamInStream->pMemPool);
    SR_BamListReset(&(pBamInStream->pAlgnLists[CURR_BIN]), pBamInStream->pMemPool);
}
Beispiel #8
0
void
bam_streamer::
set_new_region(const int ref, const int beg, const int end)
{

    if (NULL != _biter) bam_iter_destroy(_biter);

    _load_index();

    if (ref < 0)
    {
        log_os << "ERROR: Invalid region specified for BAM file: " << name() << "\n";
        exit(EXIT_FAILURE);
    }

    _biter = bam_iter_query(_bidx,ref,beg,end);
    _is_region = true;
    _region.clear();

    _is_record_set = false;
    _record_no = 0;
}
Beispiel #9
0
static int mpileup(mplp_conf_t *conf, int n, char **fn)
{
	extern void *bcf_call_add_rg(void *rghash, const char *hdtext, const char *list);
	extern void bcf_call_del_rghash(void *rghash);
	mplp_aux_t **data;
	int i, tid, pos, *n_plp, tid0 = -1, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid = -1, max_depth, max_indel_depth;
	const bam_pileup1_t **plp;
	bam_mplp_t iter;
	bam_header_t *h = 0;
	char *ref;
	void *rghash = 0;

	bcf_callaux_t *bca = 0;
	bcf_callret1_t *bcr = 0;
	bcf_call_t bc;
	bcf_t *bp = 0;
	bcf_hdr_t *bh = 0;

	bam_sample_t *sm = 0;
	kstring_t buf;
	mplp_pileup_t gplp;

	memset(&gplp, 0, sizeof(mplp_pileup_t));
	memset(&buf, 0, sizeof(kstring_t));
	memset(&bc, 0, sizeof(bcf_call_t));
	data = calloc(n, sizeof(void*));
	plp = calloc(n, sizeof(void*));
	n_plp = calloc(n, sizeof(int*));
	sm = bam_smpl_init();

	// read the header and initialize data
	for (i = 0; i < n; ++i) {
		bam_header_t *h_tmp;
		data[i] = calloc(1, sizeof(mplp_aux_t));
		data[i]->fp = strcmp(fn[i], "-") == 0? bam_dopen(fileno(stdin), "r") : bam_open(fn[i], "r");
		data[i]->conf = conf;
		h_tmp = bam_header_read(data[i]->fp);
		data[i]->h = i? h : h_tmp; // for i==0, "h" has not been set yet
		bam_smpl_add(sm, fn[i], (conf->flag&MPLP_IGNORE_RG)? 0 : h_tmp->text);
		rghash = bcf_call_add_rg(rghash, h_tmp->text, conf->pl_list);
		if (conf->reg) {
			int beg, end;
			bam_index_t *idx;
			idx = bam_index_load(fn[i]);
			if (idx == 0) {
				fprintf(stderr, "[%s] fail to load index for %d-th input.\n", __func__, i+1);
				exit(1);
			}
			if (bam_parse_region(h_tmp, conf->reg, &tid, &beg, &end) < 0) {
				fprintf(stderr, "[%s] malformatted region or wrong seqname for %d-th input.\n", __func__, i+1);
				exit(1);
			}
			if (i == 0) tid0 = tid, beg0 = beg, end0 = end;
			data[i]->iter = bam_iter_query(idx, tid, beg, end);
			bam_index_destroy(idx);
		}
		if (i == 0) h = h_tmp;
		else {
			// FIXME: to check consistency
			bam_header_destroy(h_tmp);
		}
	}
	gplp.n = sm->n;
	gplp.n_plp = calloc(sm->n, sizeof(int));
	gplp.m_plp = calloc(sm->n, sizeof(int));
	gplp.plp = calloc(sm->n, sizeof(void*));

	fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, sm->n, n);
	// write the VCF header
	if (conf->flag & MPLP_GLF) {
		kstring_t s;
		bh = calloc(1, sizeof(bcf_hdr_t));
		s.l = s.m = 0; s.s = 0;
		bp = bcf_open("-", (conf->flag&MPLP_NO_COMP)? "wu" : "w");
		for (i = 0; i < h->n_targets; ++i) {
			kputs(h->target_name[i], &s);
			kputc('\0', &s);
		}
		bh->l_nm = s.l;
		bh->name = malloc(s.l);
		memcpy(bh->name, s.s, s.l);
		s.l = 0;
		for (i = 0; i < sm->n; ++i) {
			kputs(sm->smpl[i], &s); kputc('\0', &s);
		}
		bh->l_smpl = s.l;
		bh->sname = malloc(s.l);
		memcpy(bh->sname, s.s, s.l);
		bh->txt = malloc(strlen(BAM_VERSION) + 64);
		bh->l_txt = 1 + sprintf(bh->txt, "##samtoolsVersion=%s\n", BAM_VERSION);
		free(s.s);
		bcf_hdr_sync(bh);
		bcf_hdr_write(bp, bh);
		bca = bcf_call_init(-1., conf->min_baseQ);
		bcr = calloc(sm->n, sizeof(bcf_callret1_t));
		bca->rghash = rghash;
		bca->openQ = conf->openQ, bca->extQ = conf->extQ, bca->tandemQ = conf->tandemQ;
		bca->min_frac = conf->min_frac;
		bca->min_support = conf->min_support;
	}
	if (tid0 >= 0 && conf->fai) { // region is set
		ref = faidx_fetch_seq(conf->fai, h->target_name[tid0], 0, 0x7fffffff, &ref_len);
		ref_tid = tid0;
		for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid0;
	} else ref_tid = -1, ref = 0;
	iter = bam_mplp_init(n, mplp_func, (void**)data);
	max_depth = conf->max_depth;
	if (max_depth * sm->n > 1<<20)
		fprintf(stderr, "(%s) Max depth is above 1M. Potential memory hog!\n", __func__);
	if (max_depth * sm->n < 8000) {
		max_depth = 8000 / sm->n;
		fprintf(stderr, "<%s> Set max per-file depth to %d\n", __func__, max_depth);
	}
	max_indel_depth = conf->max_indel_depth * sm->n;
	bam_mplp_set_maxcnt(iter, max_depth);
	while (bam_mplp_auto(iter, &tid, &pos, n_plp, plp) > 0) {
		if (conf->reg && (pos < beg0 || pos >= end0)) continue; // out of the region requested
		if (conf->bed && tid >= 0 && !bed_overlap(conf->bed, h->target_name[tid], pos, pos+1)) continue;
		if (tid != ref_tid) {
			free(ref); ref = 0;
			if (conf->fai) ref = faidx_fetch_seq(conf->fai, h->target_name[tid], 0, 0x7fffffff, &ref_len);
			for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid;
			ref_tid = tid;
		}
		if (conf->flag & MPLP_GLF) {
			int total_depth, _ref0, ref16;
			bcf1_t *b = calloc(1, sizeof(bcf1_t));
			for (i = total_depth = 0; i < n; ++i) total_depth += n_plp[i];
			group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp, conf->flag & MPLP_IGNORE_RG);
			_ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
			ref16 = bam_nt16_table[_ref0];
			for (i = 0; i < gplp.n; ++i)
				bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], ref16, bca, bcr + i);
			bcf_call_combine(gplp.n, bcr, ref16, &bc);
			bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
						 (conf->flag&MPLP_FMT_SP), 0, 0);
			bcf_write(bp, bh, b);
			bcf_destroy(b);
			// call indels
			if (!(conf->flag&MPLP_NO_INDEL) && total_depth < max_indel_depth && bcf_call_gap_prep(gplp.n, gplp.n_plp, gplp.plp, pos, bca, ref, rghash) >= 0) {
				for (i = 0; i < gplp.n; ++i)
					bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], -1, bca, bcr + i);
				if (bcf_call_combine(gplp.n, bcr, -1, &bc) >= 0) {
					b = calloc(1, sizeof(bcf1_t));
					bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
								 (conf->flag&MPLP_FMT_SP), bca, ref);
					bcf_write(bp, bh, b);
					bcf_destroy(b);
				}
			}
		} else {
			printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
			for (i = 0; i < n; ++i) {
				int j;
				printf("\t%d\t", n_plp[i]);
				if (n_plp[i] == 0) {
					printf("*\t*"); // FIXME: printf() is very slow...
					if (conf->flag & MPLP_PRINT_POS) printf("\t*");
				} else {
					for (j = 0; j < n_plp[i]; ++j)
						pileup_seq(plp[i] + j, pos, ref_len, ref);
					putchar('\t');
					for (j = 0; j < n_plp[i]; ++j) {
						const bam_pileup1_t *p = plp[i] + j;
						int c = bam1_qual(p->b)[p->qpos] + 33;
						if (c > 126) c = 126;
						putchar(c);
					}
					if (conf->flag & MPLP_PRINT_MAPQ) {
						putchar('\t');
						for (j = 0; j < n_plp[i]; ++j) {
							int c = plp[i][j].b->core.qual + 33;
							if (c > 126) c = 126;
							putchar(c);
						}
					}
					if (conf->flag & MPLP_PRINT_POS) {
						putchar('\t');
						for (j = 0; j < n_plp[i]; ++j) {
							if (j > 0) putchar(',');
							printf("%d", plp[i][j].qpos + 1); // FIXME: printf() is very slow...
						}
					}
				}
			}
			putchar('\n');
		}
	}

	bcf_close(bp);
	bam_smpl_destroy(sm); free(buf.s);
	for (i = 0; i < gplp.n; ++i) free(gplp.plp[i]);
	free(gplp.plp); free(gplp.n_plp); free(gplp.m_plp);
	bcf_call_del_rghash(rghash);
	bcf_hdr_destroy(bh); bcf_call_destroy(bca); free(bc.PL); free(bcr);
	bam_mplp_destroy(iter);
	bam_header_destroy(h);
	for (i = 0; i < n; ++i) {
		bam_close(data[i]->fp);
		if (data[i]->iter) bam_iter_destroy(data[i]->iter);
		free(data[i]);
	}
	free(data); free(plp); free(ref); free(n_plp);
	return 0;
}
Beispiel #10
0
int main_depth(int argc, char *argv[])
#endif
{
	int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0;
	const bam_pileup1_t **plp;
	char *reg = 0; // specified region
	void *bed = 0; // BED data structure
	bam_header_t *h = 0; // BAM header of the 1st input
	aux_t **data;
	bam_mplp_t mplp;

	// parse the command line
	while ((n = getopt(argc, argv, "r:b:q:Q:")) >= 0) {
		switch (n) {
			case 'r': reg = strdup(optarg); break;   // parsing a region requires a BAM header
			case 'b': bed = bed_read(optarg); break; // BED or position list file can be parsed now
			case 'q': baseQ = atoi(optarg); break;   // base quality threshold
			case 'Q': mapQ = atoi(optarg); break;    // mapping quality threshold
		}
	}
	if (optind == argc) {
		fprintf(stderr, "Usage: bam2depth [-r reg] [-q baseQthres] [-Q mapQthres] [-b in.bed] <in1.bam> [...]\n");
		return 1;
	}

	// initialize the auxiliary data structures
	n = argc - optind; // the number of BAMs on the command line
	data = (aux_t **) calloc(n, sizeof(void*)); // data[i] for the i-th input
	beg = 0; end = 1<<30; tid = -1;  // set the default region
	for (i = 0; i < n; ++i) {
		bam_header_t *htmp;
		data[i] = (aux_t *) calloc(1, sizeof(aux_t));
		data[i]->fp = bam_open(argv[optind+i], "r"); // open BAM
		data[i]->min_mapQ = mapQ;                    // set the mapQ filter
		htmp = bam_header_read(data[i]->fp);         // read the BAM header
		if (i == 0) {
			h = htmp; // keep the header of the 1st BAM
			if (reg) bam_parse_region(h, reg, &tid, &beg, &end); // also parse the region
		} else bam_header_destroy(htmp); // if not the 1st BAM, trash the header
		if (tid >= 0) { // if a region is specified and parsed successfully
			bam_index_t *idx = bam_index_load(argv[optind+i]);  // load the index
			data[i]->iter = bam_iter_query(idx, tid, beg, end); // set the iterator
			bam_index_destroy(idx); // the index is not needed any more; phase out of the memory
		}
	}

	// the core multi-pileup loop
	mplp = bam_mplp_init(n, read_bam, (void**)data); // initialization
	n_plp = (int*) calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
	plp = (bam_pileup1_t **) calloc(n, sizeof(void*)); // plp[i] points to the array of covering reads (internal in mplp)
	while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0) { // come to the next covered position
		if (pos < beg || pos >= end) continue; // out of range; skip
		if (bed && bed_overlap(bed, h->target_name[tid], pos, pos + 1) == 0) continue; // not in BED; skip
		fputs(h->target_name[tid], stdout); printf("\t%d", pos+1); // a customized printf() would be faster
		for (i = 0; i < n; ++i) { // base level filters have to go here
			int j, m = 0;
			for (j = 0; j < n_plp[i]; ++j) {
				const bam_pileup1_t *p = plp[i] + j; // DON'T modfity plp[][] unless you really know
				if (p->is_del || p->is_refskip) ++m; // having dels or refskips at tid:pos
				else if (bam1_qual(p->b)[p->qpos] < baseQ) ++m; // low base quality
			}
			printf("\t%d", n_plp[i] - m); // this the depth to output
		}
		putchar('\n');
	}
	free(n_plp); free(plp);
	bam_mplp_destroy(mplp);

	bam_header_destroy(h);
	for (i = 0; i < n; ++i) {
		bam_close(data[i]->fp);
		if (data[i]->iter) bam_iter_destroy(data[i]->iter);
		free(data[i]);
	}
	free(data); free(reg);
	if (bed) bed_destroy(bed);
	return 0;
}
Beispiel #11
0
int main_depth(int argc, char *argv[])
#endif
{
	int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0, min_len = 0, nfiles;
	const bam_pileup1_t **plp;
	char *reg = 0; // specified region
	void *bed = 0; // BED data structure
    char *file_list = NULL, **fn = NULL;
	bam_header_t *h = 0; // BAM header of the 1st input
	aux_t **data;
	bam_mplp_t mplp;

	// parse the command line
	while ((n = getopt(argc, argv, "r:b:q:Q:l:f:")) >= 0) {
		switch (n) {
			case 'l': min_len = atoi(optarg); break; // minimum query length
			case 'r': reg = strdup(optarg); break;   // parsing a region requires a BAM header
			case 'b': bed = bed_read(optarg); break; // BED or position list file can be parsed now
			case 'q': baseQ = atoi(optarg); break;   // base quality threshold
			case 'Q': mapQ = atoi(optarg); break;    // mapping quality threshold
			case 'f': file_list = optarg; break;
		}
	}
	if (optind == argc && !file_list) {
        fprintf(stderr, "\n");
        fprintf(stderr, "Usage: samtools depth [options] in1.bam [in2.bam [...]]\n");
        fprintf(stderr, "Options:\n");
        fprintf(stderr, "   -b <bed>            list of positions or regions\n");
        fprintf(stderr, "   -f <list>           list of input BAM filenames, one per line [null]\n");
        fprintf(stderr, "   -l <int>            minQLen\n");
        fprintf(stderr, "   -q <int>            base quality threshold\n");
        fprintf(stderr, "   -Q <int>            mapping quality threshold\n");
        fprintf(stderr, "   -r <chr:from-to>    region\n");
        fprintf(stderr, "\n");
		return 1;
	}

	// initialize the auxiliary data structures
    if (file_list) 
    {
        if ( read_file_list(file_list,&nfiles,&fn) ) return 1;
        n = nfiles;
        argv = fn;
        optind = 0;
    }
    else
        n = argc - optind; // the number of BAMs on the command line
	data = calloc(n, sizeof(void*)); // data[i] for the i-th input
	beg = 0; end = 1<<30; tid = -1;  // set the default region
	for (i = 0; i < n; ++i) {
		bam_header_t *htmp;
		data[i] = calloc(1, sizeof(aux_t));
		data[i]->fp = bam_open(argv[optind+i], "r"); // open BAM
		data[i]->min_mapQ = mapQ;                    // set the mapQ filter
		data[i]->min_len  = min_len;                 // set the qlen filter
		htmp = bam_header_read(data[i]->fp);         // read the BAM header
		if (i == 0) {
			h = htmp; // keep the header of the 1st BAM
			if (reg) bam_parse_region(h, reg, &tid, &beg, &end); // also parse the region
		} else bam_header_destroy(htmp); // if not the 1st BAM, trash the header
		if (tid >= 0) { // if a region is specified and parsed successfully
			bam_index_t *idx = bam_index_load(argv[optind+i]);  // load the index
			data[i]->iter = bam_iter_query(idx, tid, beg, end); // set the iterator
			bam_index_destroy(idx); // the index is not needed any more; phase out of the memory
		}
	}

	// the core multi-pileup loop
	mplp = bam_mplp_init(n, read_bam, (void**)data); // initialization
	bam_mplp_set_maxcnt(mplp,2147483647); // set max_depth to int max
	n_plp = calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
	plp = calloc(n, sizeof(void*)); // plp[i] points to the array of covering reads (internal in mplp)
	while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0) { // come to the next covered position
		if (pos < beg || pos >= end) continue; // out of range; skip
		if (bed && bed_overlap(bed, h->target_name[tid], pos, pos + 1) == 0) continue; // not in BED; skip
		fputs(h->target_name[tid], stdout); printf("\t%d", pos+1); // a customized printf() would be faster
		for (i = 0; i < n; ++i) { // base level filters have to go here
			int j, m = 0;
			for (j = 0; j < n_plp[i]; ++j) {
				const bam_pileup1_t *p = plp[i] + j; // DON'T modfity plp[][] unless you really know
				if (p->is_del || p->is_refskip) ++m; // having dels or refskips at tid:pos
				else if (bam1_qual(p->b)[p->qpos] < baseQ) ++m; // low base quality
			}
			printf("\t%d", n_plp[i] - m); // this the depth to output
		}
		putchar('\n');
	}
	free(n_plp); free(plp);
	bam_mplp_destroy(mplp);

	bam_header_destroy(h);
	for (i = 0; i < n; ++i) {
		bam_close(data[i]->fp);
		if (data[i]->iter) bam_iter_destroy(data[i]->iter);
		free(data[i]);
	}
	free(data); free(reg);
	if (bed) bed_destroy(bed);
    if ( file_list )
    {
        for (i=0; i<n; i++) free(fn[i]);
        free(fn);
    }
	return 0;
}
Beispiel #12
0
int bam_merge_core2(int by_qname, const char *out, const char *headers, int n, char * const *fn, int flag, const char *reg, int level)
#endif
{
	bamFile fpout, *fp;
	heap1_t *heap;
	bam_header_t *hout = 0;
	bam_header_t *hheaders = NULL;
	int i, j, *RG_len = 0;
	uint64_t idx = 0;
	char **RG = 0, mode[8];
	bam_iter_t *iter = 0;

	if (headers) {
		tamFile fpheaders = sam_open(headers);
		if (fpheaders == 0) {
			const char *message = strerror(errno);
			fprintf(stderr, "[bam_merge_core] cannot open '%s': %s\n", headers, message);
			return -1;
		}
		hheaders = sam_header_read(fpheaders);
		sam_close(fpheaders);
	}

	g_is_by_qname = by_qname;
	fp = (bamFile*)calloc(n, sizeof(bamFile));
	heap = (heap1_t*)calloc(n, sizeof(heap1_t));
	iter = (bam_iter_t*)calloc(n, sizeof(bam_iter_t));
	// prepare RG tag
	if (flag & MERGE_RG) {
		RG = (char**)calloc(n, sizeof(void*));
		RG_len = (int*)calloc(n, sizeof(int));
		for (i = 0; i != n; ++i) {
			int l = strlen(fn[i]);
			const char *s = fn[i];
			if (l > 4 && strcmp(s + l - 4, ".bam") == 0) l -= 4;
			for (j = l - 1; j >= 0; --j) if (s[j] == '/') break;
			++j; l -= j;
			RG[i] = calloc(l + 1, 1);
			RG_len[i] = l;
			strncpy(RG[i], s + j, l);
		}
	}
	// read the first
	for (i = 0; i != n; ++i) {
		bam_header_t *hin;
		fp[i] = bam_open(fn[i], "r");
		if (fp[i] == 0) {
			int j;
			fprintf(stderr, "[bam_merge_core] fail to open file %s\n", fn[i]);
			for (j = 0; j < i; ++j) bam_close(fp[j]);
			free(fp); free(heap);
			// FIXME: possible memory leak
			return -1;
		}
		hin = bam_header_read(fp[i]);
		if (i == 0) { // the first BAM
			hout = hin;
		} else { // validate multiple baf
			int min_n_targets = hout->n_targets;
			if (hin->n_targets < min_n_targets) min_n_targets = hin->n_targets;

			for (j = 0; j < min_n_targets; ++j)
				if (strcmp(hout->target_name[j], hin->target_name[j]) != 0) {
					fprintf(stderr, "[bam_merge_core] different target sequence name: '%s' != '%s' in file '%s'\n",
							hout->target_name[j], hin->target_name[j], fn[i]);
					return -1;
				}

			// If this input file has additional target reference sequences,
			// add them to the headers to be output
			if (hin->n_targets > hout->n_targets) {
				swap_header_targets(hout, hin);
				// FIXME Possibly we should also create @SQ text headers
				// for the newly added reference sequences
			}

			bam_header_destroy(hin);
		}
	}

	if (hheaders) {
		// If the text headers to be swapped in include any @SQ headers,
		// check that they are consistent with the existing binary list
		// of reference information.
		if (hheaders->n_targets > 0) {
			if (hout->n_targets != hheaders->n_targets) {
				fprintf(stderr, "[bam_merge_core] number of @SQ headers in '%s' differs from number of target sequences\n", headers);
				if (!reg) return -1;
			}
			for (j = 0; j < hout->n_targets; ++j)
				if (strcmp(hout->target_name[j], hheaders->target_name[j]) != 0) {
					fprintf(stderr, "[bam_merge_core] @SQ header '%s' in '%s' differs from target sequence\n", hheaders->target_name[j], headers);
					if (!reg) return -1;
				}
		}

		swap_header_text(hout, hheaders);
		bam_header_destroy(hheaders);
	}

	if (reg) {
		int tid, beg, end;
		if (bam_parse_region(hout, reg, &tid, &beg, &end) < 0) {
			fprintf(stderr, "[%s] Malformated region string or undefined reference name\n", __func__);
			return -1;
		}
		for (i = 0; i < n; ++i) {
			bam_index_t *idx;
			idx = bam_index_load(fn[i]);
			iter[i] = bam_iter_query(idx, tid, beg, end);
			bam_index_destroy(idx);
		}
	}

	for (i = 0; i < n; ++i) {
		heap1_t *h = heap + i;
		h->i = i;
		h->b = (bam1_t*)calloc(1, sizeof(bam1_t));
		if (bam_iter_read(fp[i], iter[i], h->b) >= 0) {
			h->pos = ((uint64_t)h->b->core.tid<<32) | (uint32_t)((int32_t)h->b->core.pos+1)<<1 | bam1_strand(h->b);
			h->idx = idx++;
		}
		else h->pos = HEAP_EMPTY;
	}
	if (flag & MERGE_UNCOMP) level = 0;
	else if (flag & MERGE_LEVEL1) level = 1;
	strcpy(mode, "w");
	if (level >= 0) sprintf(mode + 1, "%d", level < 9? level : 9);
	if ((fpout = strcmp(out, "-")? bam_open(out, "w") : bam_dopen(fileno(stdout), "w")) == 0) {
		fprintf(stderr, "[%s] fail to create the output file.\n", __func__);
		return -1;
	}
	bam_header_write(fpout, hout);
	bam_header_destroy(hout);
#ifndef _PBGZF_USE 
	if (!(flag & MERGE_UNCOMP)) bgzf_mt(fpout, n_threads, 256);
#endif

	ks_heapmake(heap, n, heap);
	while (heap->pos != HEAP_EMPTY) {
		bam1_t *b = heap->b;
		if (flag & MERGE_RG) {
			uint8_t *rg = bam_aux_get(b, "RG");
			if (rg) bam_aux_del(b, rg);
			bam_aux_append(b, "RG", 'Z', RG_len[heap->i] + 1, (uint8_t*)RG[heap->i]);
		}
		bam_write1_core(fpout, &b->core, b->data_len, b->data);
		if ((j = bam_iter_read(fp[heap->i], iter[heap->i], b)) >= 0) {
			heap->pos = ((uint64_t)b->core.tid<<32) | (uint32_t)((int)b->core.pos+1)<<1 | bam1_strand(b);
			heap->idx = idx++;
		} else if (j == -1) {
			heap->pos = HEAP_EMPTY;
			free(heap->b->data); free(heap->b);
			heap->b = 0;
		} else fprintf(stderr, "[bam_merge_core] '%s' is truncated. Continue anyway.\n", fn[heap->i]);
		ks_heapadjust(heap, 0, n, heap);
	}

	if (flag & MERGE_RG) {
		for (i = 0; i != n; ++i) free(RG[i]);
		free(RG); free(RG_len);
	}
	for (i = 0; i != n; ++i) {
		bam_iter_destroy(iter[i]);
		bam_close(fp[i]);
	}
	bam_close(fpout);
	free(fp); free(heap); free(iter);
	return 0;
}
Beispiel #13
0
static int mpileup(mplp_conf_t *conf, int n, char **fn)
{
	extern void *bcf_call_add_rg(void *rghash, const char *hdtext, const char *list);
	extern void bcf_call_del_rghash(void *rghash);
	mplp_aux_t **data;
	int i, tid, pos, *n_plp, tid0 = -1, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid = -1, max_depth, max_indel_depth;
	const bam_pileup1_t **plp;
	bam_mplp_t iter;
	bam_header_t *h = 0;
	char *ref;
	void *rghash = 0;

	bcf_callaux_t *bca = 0;
	bcf_callret1_t *bcr = 0;
	bcf_call_t bc;
	bcf_t *bp = 0;
	bcf_hdr_t *bh = 0;

	bam_sample_t *sm = 0;
	kstring_t buf;
	mplp_pileup_t gplp;

	memset(&gplp, 0, sizeof(mplp_pileup_t));
	memset(&buf, 0, sizeof(kstring_t));
	memset(&bc, 0, sizeof(bcf_call_t));
	data = calloc(n, sizeof(void*));
	plp = calloc(n, sizeof(void*));
	n_plp = calloc(n, sizeof(int*));
	sm = bam_smpl_init();

	// read the header and initialize data
	for (i = 0; i < n; ++i) {
		bam_header_t *h_tmp;
		data[i] = calloc(1, sizeof(mplp_aux_t));
		data[i]->fp = strcmp(fn[i], "-") == 0? bam_dopen(fileno(stdin), "r") : bam_open(fn[i], "r");
		data[i]->conf = conf;
		h_tmp = bam_header_read(data[i]->fp);
		data[i]->h = i? h : h_tmp; // for i==0, "h" has not been set yet
		bam_smpl_add(sm, fn[i], (conf->flag&MPLP_IGNORE_RG)? 0 : h_tmp->text);
		rghash = bcf_call_add_rg(rghash, h_tmp->text, conf->pl_list);
		if (conf->reg) {
			int beg, end;
			bam_index_t *idx;
			idx = bam_index_load(fn[i]);
			if (idx == 0) {
				fprintf(stderr, "[%s] fail to load index for %d-th input.\n", __func__, i+1);
				exit(1);
			}
			if (bam_parse_region(h_tmp, conf->reg, &tid, &beg, &end) < 0) {
				fprintf(stderr, "[%s] malformatted region or wrong seqname for %d-th input.\n", __func__, i+1);
				exit(1);
			}
			if (i == 0) tid0 = tid, beg0 = beg, end0 = end;
			data[i]->iter = bam_iter_query(idx, tid, beg, end);
			bam_index_destroy(idx);
		}
		if (i == 0) h = h_tmp;
		else {
			// FIXME: to check consistency
			bam_header_destroy(h_tmp);
		}
	}
	gplp.n = sm->n;
	gplp.n_plp = calloc(sm->n, sizeof(int));
	gplp.m_plp = calloc(sm->n, sizeof(int));
	gplp.plp = calloc(sm->n, sizeof(void*));

	fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, sm->n, n);
	// write the VCF header
	if (conf->flag & MPLP_GLF) {
		kstring_t s;
		bh = calloc(1, sizeof(bcf_hdr_t));
		s.l = s.m = 0; s.s = 0;
		bp = bcf_open("-", (conf->flag&MPLP_NO_COMP)? "wu" : "w");
		for (i = 0; i < h->n_targets; ++i) {
			kputs(h->target_name[i], &s);
			kputc('\0', &s);
		}
		bh->l_nm = s.l;
		bh->name = malloc(s.l);
		memcpy(bh->name, s.s, s.l);
		s.l = 0;
		for (i = 0; i < sm->n; ++i) {
			kputs(sm->smpl[i], &s); kputc('\0', &s);
		}
		bh->l_smpl = s.l;
		bh->sname = malloc(s.l);
		memcpy(bh->sname, s.s, s.l);
		bh->txt = malloc(strlen(BAM_VERSION) + 64);
		bh->l_txt = 1 + sprintf(bh->txt, "##samtoolsVersion=%s\n", BAM_VERSION);
		free(s.s);
		bcf_hdr_sync(bh);
		bcf_hdr_write(bp, bh);
		bca = bcf_call_init(-1., conf->min_baseQ);
		bcr = calloc(sm->n, sizeof(bcf_callret1_t));
		bca->rghash = rghash;
		bca->openQ = conf->openQ, bca->extQ = conf->extQ, bca->tandemQ = conf->tandemQ;
		bca->min_frac = conf->min_frac;
		bca->min_support = conf->min_support;
	}
	if (tid0 >= 0 && conf->fai) { // region is set
		ref = faidx_fetch_seq(conf->fai, h->target_name[tid0], 0, 0x7fffffff, &ref_len);
		ref_tid = tid0;
		for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid0;
	} else ref_tid = -1, ref = 0;
	iter = bam_mplp_init(n, mplp_func, (void**)data);
	max_depth = conf->max_depth;
	if (max_depth * sm->n > 1<<20)
		fprintf(stderr, "(%s) Max depth is above 1M. Potential memory hog!\n", __func__);
	if (max_depth * sm->n < 8000) {
		max_depth = 8000 / sm->n;
		fprintf(stderr, "<%s> Set max per-file depth to %d\n", __func__, max_depth);
	}
	max_indel_depth = conf->max_indel_depth * sm->n;
	bam_mplp_set_maxcnt(iter, max_depth);


	int storeSize = 100;

	int delStore[2][100] = {{0},{0}};

	typedef char * mstring;

	while (bam_mplp_auto(iter, &tid, &pos, n_plp, plp) > 0) {
		if (conf->reg && (pos < beg0 || pos >= end0)) continue; // out of the region requested
		if (conf->bed && tid >= 0 && !bed_overlap(conf->bed, h->target_name[tid], pos, pos+1)) continue;
		if (tid != ref_tid) {
			free(ref); ref = 0;
			if (conf->fai) ref = faidx_fetch_seq(conf->fai, h->target_name[tid], 0, 0x7fffffff, &ref_len);
			for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid;
			ref_tid = tid;
		}
		if (conf->flag & MPLP_GLF) {
			int total_depth, _ref0, ref16;
			bcf1_t *b = calloc(1, sizeof(bcf1_t));
			for (i = total_depth = 0; i < n; ++i) total_depth += n_plp[i];
			group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp, conf->flag & MPLP_IGNORE_RG);
			_ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
			ref16 = bam_nt16_table[_ref0];
			for (i = 0; i < gplp.n; ++i)
				bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], ref16, bca, bcr + i);
			bcf_call_combine(gplp.n, bcr, ref16, &bc);
			bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
						 (conf->flag&MPLP_FMT_SP), 0, 0);
			bcf_write(bp, bh, b);
			bcf_destroy(b);
			// call indels
			if (!(conf->flag&MPLP_NO_INDEL) && total_depth < max_indel_depth && bcf_call_gap_prep(gplp.n, gplp.n_plp, gplp.plp, pos, bca, ref, rghash) >= 0) {
				for (i = 0; i < gplp.n; ++i)
					bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], -1, bca, bcr + i);
				if (bcf_call_combine(gplp.n, bcr, -1, &bc) >= 0) {
					b = calloc(1, sizeof(bcf1_t));
					bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
								 (conf->flag&MPLP_FMT_SP), bca, ref);
					bcf_write(bp, bh, b);
					bcf_destroy(b);
				}
			}
		} else {
			printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
			for (i = 0; i < n; ++i) {
				int j;
				printf("\t%d\t", n_plp[i]);
				if (n_plp[i] == 0) {
					printf("*\t*"); // FIXME: printf() is very slow...
					if (conf->flag & MPLP_PRINT_POS) printf("\t*");
				} else {
					//MDW start					
					//for each position in the pileup column
					int charLen = 16;
					int countChars[ charLen ][2];
					int countiChars[ charLen ][2];

					int countGap[2]={0,0};

					//double qvTotal=0;
					int numStruck=0;
					int numGood=0;
					int tti;
					int ttj;
					mstring insAllele[100];
					int insAlleleCnt[100];
					int sf=0;
					int flag=0;

					//typedef char * string;
					char insStr0[10000];
					int iCnt0=0;

					char insStr1[10000];
					int iCnt1=0;

					char delStr0[10000];
					int dCnt0=0;

					char delStr1[10000];
					int dCnt1=0;


					float qposP[10000];
					int qposCnt=0;



					//initialize with zeros
						for(tti=0;tti<charLen;tti++){
						  countChars[tti][0]=0;
						  countChars[tti][1]=0;
						}

					// define repeat length here; look back up to 10 prior positions
					// start one position away.
					int replC=0; //
					for(tti=1;tti<=15;tti++){
						// check for greater than zero
						if(toupper(ref[pos-1])==toupper(ref[pos-tti])){
							replC++;
						}else{ // breaks the chain at first non identical to current position not strict homopolymer
							break;
						}
					}					
					int reprC=0; // 
					for(tti=1;tti<=15;tti++){
						// check for greater than zero
						if(toupper(ref[pos+1])==toupper(ref[pos+tti])){
							reprC++;
						}else{ // breaks the chain at first non identical to current position not strict homopolymer
							break;
						}
					}		
					int repT = replC;
					if(replC < reprC){
						repT=reprC;
					}



					for (j = 0; j < n_plp[i]; ++j){
						const bam_pileup1_t *p = plp[i] + j;
									
						/*
						SAME LOGIC AS pileup_seq()
						*/

						if(p->is_refskip){ // never count intron gaps in numStruck
							continue;
						}

						if(p->is_del){ // skip deletion gap, after first position which is the first aligned char
							continue;
						}

						if( 	p->b->core.qual < conf->min_mqToCount  || // mapping quality
							conf->maxrepC < (repT) || // max homopolymer run, this will not 
							(!p->is_del && bam1_qual(p->b)[p->qpos] < conf->min_baseQ) || // base quality for matches
							p->alignedQPosBeg <= (conf->trimEnd ) || p->alignedQPosEnd <= (conf->trimEnd ) ||  // trimEnd is 1-based
							p->zf == 1 || // fusion tag
							p->ih > conf->maxIH  || // max hit index
							(p->nmd > conf->maxNM) || // max mismatch
							(conf->flagFilter == 1 && !(p->b->core.flag&BAM_FPROPER_PAIR)) || // optionally keep only proper pairs
							(conf->flagFilter == 2 && p->b->core.flag&BAM_FSECONDARY) || // optionally strike secondary
							(conf->flagFilter == 3 && p->b->core.flag&BAM_FDUP) || // optionally strike dup
							(conf->flagFilter == 4 && (p->b->core.flag&BAM_FDUP || p->b->core.flag&BAM_FSECONDARY))  || // optionally strike secondary or dup
							(conf->flagFilter == 5 && (p->b->core.flag&BAM_FDUP || p->b->core.flag&BAM_FSECONDARY || p->b->core.flag&BAM_FQCFAIL || !(p->b->core.flag&BAM_FPROPER_PAIR) ))   // optionally strike secondary, dup and QCfail


						){
							numStruck++;
							continue;
						}

						
						//printf("repT=%d: %d %c %c %c %c \n",repT,p->indel,ref[pos],ref[pos-1],ref[pos-2],ref[pos-3]);


						if(!p->is_del && p->indel==0){
  						  countChars[ bam1_seqi(bam1_seq(p->b), p->qpos) ][ bam1_strand(p->b) ] ++;
						  numGood++;			

						}else if(p->is_refskip){
						  countGap[ bam1_strand(p->b) ]++;
						}
						
						if(p->indel<0){
    						  numGood++;			
						  if(bam1_strand(p->b) ==0){
							  for(tti=1;tti<= -p->indel; tti++) {
							    // current spot, starting at 0 in store, because indel<0 refers to next position
							   delStr0[dCnt0] =  ref[pos+tti];
							   dCnt0++;
							  }	
							  delStr0[dCnt0] = ',';
							  dCnt0++;
						  }else{
							  for(tti=1;tti<= -p->indel; tti++) {
							    // current spot, starting at 0 in store, because indel<0 refers to next position
							   delStr1[dCnt1] = ref[pos+tti];
							   dCnt1++;
							  }	
							  delStr1[dCnt1] = ',';
							  dCnt1++;
						  }



						}else if(p->indel>0){
						  numGood++;			

						  if(bam1_strand(p->b) ==0){
							  for(tti=1;tti<= p->indel; tti++) {
							    // current spot, starting at 0 in store, because indel<0 refers to next position
							   insStr0[iCnt0] = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + tti)];
							   iCnt0++;
							  }	
							  insStr0[iCnt0] = ',';
							  iCnt0++;
						  }else{
							  for(tti=1;tti<= p->indel; tti++) {
							    // current spot, starting at 0 in store, because indel<0 refers to next position
							   insStr1[iCnt1] = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + tti)];
							   iCnt1++;
							  }	
							  insStr1[iCnt1] = ',';
							  iCnt1++;
						  }


						}
						//calculate position of variant within aligned read - no soft clips
						if( toupper(ref[pos]) != toupper(bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)]) || p->indel>0 || p->indel<0  ){

						//distance to end; calculate distance to end of aligned read.  removes soft clips.
						int distToEnd = (p->alignedQPosBeg < p->alignedQPosEnd) ? p->alignedQPosBeg : p->alignedQPosEnd;
						qposP[qposCnt] = distToEnd;						  
						qposCnt++;	
						// printf("id=%s, pos=%d",bam1_qname(p->b),distToEnd);
						}	
					}

					//

					//print A,C,G,T, by +/-
				        printf("\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d", 	countChars[1][0],countChars[1][1],
											countChars[2][0],countChars[2][1],
											countChars[4][0],countChars[4][1],
											countChars[8][0],countChars[8][1],
											countChars[7][0],countChars[7][1]);
					
					putchar('\t');
					for(tti=0;tti<dCnt0;tti++){
					  putchar(delStr0[tti]);
					}

					putchar('\t');
					for(tti=0;tti<dCnt1;tti++){
					  putchar(delStr1[tti]);
					}

					putchar('\t');
					for(tti=0;tti<iCnt0;tti++){
					  putchar(insStr0[tti]);
					}

					putchar('\t');
					for(tti=0;tti<iCnt1;tti++){
					  putchar(insStr1[tti]);
					}

					printf("\t%d\t%d",numGood,numStruck);					

					// get non-ref qpos variation

					float medqpos = -1;
					float medAbsDev = -1;
					if(qposCnt>0){
					  medqpos = median(qposCnt,qposP);
					  float absDev[qposCnt];
					  for(tti=0;tti<qposCnt;tti++){
						absDev[tti] = abs(medqpos - qposP[tti]);
					  }
					  medAbsDev = median(qposCnt-1,absDev);
					}
					printf("\t%f",medAbsDev);

					///END MDW
				}



			}
			putchar('\n');
		}
	}

	bcf_close(bp);
	bam_smpl_destroy(sm); free(buf.s);
	for (i = 0; i < gplp.n; ++i) free(gplp.plp[i]);
	free(gplp.plp); free(gplp.n_plp); free(gplp.m_plp);
	bcf_call_del_rghash(rghash);
	bcf_hdr_destroy(bh); bcf_call_destroy(bca); free(bc.PL); free(bcr);
	bam_mplp_destroy(iter);
	bam_header_destroy(h);
	for (i = 0; i < n; ++i) {
		bam_close(data[i]->fp);
		if (data[i]->iter) bam_iter_destroy(data[i]->iter);
		free(data[i]);
	}
	free(data); free(plp); free(ref); free(n_plp);
	return 0;
}
Beispiel #14
0
int main_bedcov(int argc, char *argv[])
{
	extern void bam_init_header_hash(bam_header_t*);
	gzFile fp;
	kstring_t str;
	kstream_t *ks;
	bam_index_t **idx;
	bam_header_t *h = 0;
	aux_t **aux;
	int *n_plp, dret, i, n, c, min_mapQ = 0;
	int64_t *cnt;
	const bam_pileup1_t **plp;

	while ((c = getopt(argc, argv, "Q:")) >= 0) {
		switch (c) {
		case 'Q': min_mapQ = atoi(optarg); break;
		}
	}
	if (optind + 2 > argc) {
		fprintf(stderr, "Usage: samtools bedcov <in.bed> <in1.bam> [...]\n");
		return 1;
	}
	memset(&str, 0, sizeof(kstring_t));
	n = argc - optind - 1;
	aux = calloc(n, sizeof(aux_t*));
	idx = calloc(n, sizeof(bam_index_t*));
	for (i = 0; i < n; ++i) {
		aux[i] = calloc(1, sizeof(aux_t));
		aux[i]->min_mapQ = min_mapQ;
		aux[i]->fp = bam_open(argv[i+optind+1], "r");
		idx[i] = bam_index_load(argv[i+optind+1]);
		if (aux[i]->fp == 0 || idx[i] == 0) {
			fprintf(stderr, "ERROR: fail to open index BAM file '%s'\n", argv[i+optind+1]);
			return 2;
		}
		bgzf_set_cache_size(aux[i]->fp, 20);
		if (i == 0) h = bam_header_read(aux[0]->fp);
	}
	bam_init_header_hash(h);
	cnt = calloc(n, 8);

	fp = gzopen(argv[optind], "rb");
	ks = ks_init(fp);
	n_plp = calloc(n, sizeof(int));
	plp = calloc(n, sizeof(bam_pileup1_t*));
	while (ks_getuntil(ks, KS_SEP_LINE, &str, &dret) >= 0) {
		char *p, *q;
		int tid, beg, end, pos;
		bam_mplp_t mplp;

		for (p = q = str.s; *p && *p != '\t'; ++p);
		if (*p != '\t') goto bed_error;
		*p = 0; tid = bam_get_tid(h, q); *p = '\t';
		if (tid < 0) goto bed_error;
		for (q = p = p + 1; isdigit(*p); ++p);
		if (*p != '\t') goto bed_error;
		*p = 0; beg = atoi(q); *p = '\t';
		for (q = p = p + 1; isdigit(*p); ++p);
		if (*p == '\t' || *p == 0) {
			int c = *p;
			*p = 0; end = atoi(q); *p = c;
		} else goto bed_error;

		for (i = 0; i < n; ++i) {
			if (aux[i]->iter) bam_iter_destroy(aux[i]->iter);
			aux[i]->iter = bam_iter_query(idx[i], tid, beg, end);
		}
		mplp = bam_mplp_init(n, read_bam, (void**)aux);
		bam_mplp_set_maxcnt(mplp, 64000);
		memset(cnt, 0, 8 * n);
		while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0)
			if (pos >= beg && pos < end)
				for (i = 0; i < n; ++i) cnt[i] += n_plp[i];
		for (i = 0; i < n; ++i) {
			kputc('\t', &str);
			kputl(cnt[i], &str);
		}
		puts(str.s);
		bam_mplp_destroy(mplp);
		continue;

bed_error:
		fprintf(stderr, "Errors in BED line '%s'\n", str.s);
	}
	free(n_plp); free(plp);
	ks_destroy(ks);
	gzclose(fp);

	free(cnt);
	for (i = 0; i < n; ++i) {
		if (aux[i]->iter) bam_iter_destroy(aux[i]->iter);
		bam_index_destroy(idx[i]);
		bam_close(aux[i]->fp);
		free(aux[i]);
	}
	bam_header_destroy(h);
	free(aux); free(idx);
	free(str.s);
	return 0;
}
Beispiel #15
0
// jump to a certain chromosome in a bam file
SR_Status SR_BamInStreamJump(SR_BamInStream* pBamInStream, int32_t refID, int32_t begin, int32_t end)
{
    // if we do not have the index file return error
    if (pBamInStream->pBamIndex == NULL)
        return SR_ERR;
    
    if (pBamInStream->pBamIterator != NULL) {
      bam_iter_destroy(*(pBamInStream->pBamIterator));
      free(pBamInStream->pBamIterator);
      pBamInStream->pBamIterator = NULL;
    }

    // clear the bam array before jump
    SR_BamInStreamReset(pBamInStream);

    pBamInStream->pBamIterator = (bam_iter_t*) malloc(sizeof(bam_iter_t));

    // jump and read the first alignment in the given chromosome
    int ret;
    //bam_iter_t pBamIter = bam_iter_query(pBamInStream->pBamIndex, refID, begin, end);
    *pBamInStream->pBamIterator = bam_iter_query(pBamInStream->pBamIndex, refID, begin, end);

    pBamInStream->pNewNode = SR_BamNodeAlloc(pBamInStream->pMemPool);
    ret = bam_iter_read(pBamInStream->fpBamInput, *(pBamInStream->pBamIterator), &(pBamInStream->pNewNode->alignment));
    //bam_iter_destroy(pBamIter);

    khash_t(queryName)* pNameHashCurr = NULL;

    // see if we jump to the desired chromosome
    if (ret > 0 && pBamInStream->pNewNode->alignment.core.tid == refID)
    {
        // exclude those reads who are non-paired-end, qc-fail, duplicate-marked, proper-paired, 
        // both aligned, secondary-alignment and no-name-specified.
        if ((pBamInStream->pNewNode->alignment.core.flag & SR_BAM_FMASK) != 0
            || strcmp(bam1_qname(&(pBamInStream->pNewNode->alignment)), "*") == 0)
        {
            SR_BamNodeFree(pBamInStream->pNewNode, pBamInStream->pMemPool);
            pBamInStream->pNewNode = NULL;
            pBamInStream->currBinPos = NO_QUERY_YET;
        }
        else
        {
            SR_BamListPushHead(&(pBamInStream->pAlgnLists[CURR_BIN]), pBamInStream->pNewNode);

            int khRet = 0;
            khiter_t khIter = kh_put(queryName, pBamInStream->pNameHashes[CURR_BIN], bam1_qname(&(pBamInStream->pNewNode->alignment)), &khRet);

            if (khRet != 0)
            {
                pNameHashCurr = pBamInStream->pNameHashes[CURR_BIN];
                kh_value(pNameHashCurr, khIter) = pBamInStream->pNewNode;
            }
            else
                return SR_ERR;

            pBamInStream->currBinPos = pBamInStream->pNewNode->alignment.core.pos; 
            pBamInStream->pNewNode = NULL;
        }

        pBamInStream->currRefID = refID;
        return SR_OK;
    }
    else if (ret == -1)
    {
        return SR_OUT_OF_RANGE;
    }
    else
    {
        return SR_ERR;
    }
}