X-Git-Url: http://woldlab.caltech.edu/gitweb/?p=pysam.git;a=blobdiff_plain;f=samtools%2Fbam_plcmd.c.pysam.c;h=93b729db2388b9ef14ba093c2eef7b31dee52663;hp=473d409633dd80836c0ffccedb00c31fa1246e16;hb=e1756c41e7a1d7cc01fb95e42df9dd04da2d2991;hpb=ca46ef4ba4a883c57cea62d5bf1bc021f1185109

diff --git a/samtools/bam_plcmd.c.pysam.c b/samtools/bam_plcmd.c.pysam.c
index 473d409..93b729d 100644
--- a/samtools/bam_plcmd.c.pysam.c
+++ b/samtools/bam_plcmd.c.pysam.c
@@ -8,8 +8,93 @@
 #include <errno.h>
 #include "sam.h"
 #include "faidx.h"
+#include "bam_maqcns.h"
+#include "khash.h"
+#include "glf.h"
 #include "kstring.h"
 
+typedef int *indel_list_t;
+KHASH_MAP_INIT_INT64(64, indel_list_t)
+
+#define BAM_PLF_SIMPLE     0x01
+#define BAM_PLF_CNS        0x02
+#define BAM_PLF_INDEL_ONLY 0x04
+#define BAM_PLF_GLF        0x08
+#define BAM_PLF_VAR_ONLY   0x10
+#define BAM_PLF_2ND        0x20
+#define BAM_PLF_RANBASE    0x40
+#define BAM_PLF_1STBASE    0x80
+#define BAM_PLF_ALLBASE    0x100
+#define BAM_PLF_READPOS    0x200
+#define BAM_PLF_NOBAQ      0x400
+
+typedef struct {
+	bam_header_t *h;
+	bam_maqcns_t *c;
+	bam_maqindel_opt_t *ido;
+	faidx_t *fai;
+	khash_t(64) *hash;
+	uint32_t format;
+	int tid, len, last_pos;
+	int mask;
+	int capQ_thres, min_baseQ;
+    int max_depth;  // for indel calling, ignore reads with the depth too high. 0 for unlimited
+	char *ref;
+	glfFile fp_glf; // for glf output only
+} pu_data_t;
+
+char **__bam_get_lines(const char *fn, int *_n);
+void bam_init_header_hash(bam_header_t *header);
+int32_t bam_get_tid(const bam_header_t *header, const char *seq_name);
+
+static khash_t(64) *load_pos(const char *fn, bam_header_t *h)
+{
+	char **list;
+	int i, j, n, *fields, max_fields;
+	khash_t(64) *hash;
+	bam_init_header_hash(h);
+	list = __bam_get_lines(fn, &n);
+	hash = kh_init(64);
+	max_fields = 0; fields = 0;
+	for (i = 0; i < n; ++i) {
+		char *str = list[i];
+		int chr, n_fields, ret;
+		khint_t k;
+		uint64_t x;
+		n_fields = ksplit_core(str, 0, &max_fields, &fields);
+		if (n_fields < 2) continue;
+		chr = bam_get_tid(h, str + fields[0]);
+		if (chr < 0) {
+			fprintf(pysamerr, "[load_pos] unknown reference sequence name: %s\n", str + fields[0]);
+			continue;
+		}
+		x = (uint64_t)chr << 32 | (atoi(str + fields[1]) - 1);
+		k = kh_put(64, hash, x, &ret);
+		if (ret == 0) {
+			fprintf(pysamerr, "[load_pos] position %s:%s has been loaded.\n", str+fields[0], str+fields[1]);
+			continue;
+		}
+		kh_val(hash, k) = 0;
+		if (n_fields > 2) {
+			// count
+			for (j = 2; j < n_fields; ++j) {
+				char *s = str + fields[j];
+				if ((*s != '+' && *s != '-') || !isdigit(s[1])) break;
+ 			}
+			if (j > 2) { // update kh_val()
+				int *q, y, z;
+				q = kh_val(hash, k) = (int*)calloc(j - 1, sizeof(int));
+				q[0] = j - 2; z = j; y = 1;
+				for (j = 2; j < z; ++j)
+					q[y++] = atoi(str + fields[j]);
+			}
+		}
+		free(str);
+	}
+	free(list); free(fields);
+	return hash;
+}
+
 static inline int printw(int c, FILE *fp)
 {
 	char buf[16];
@@ -25,6 +110,75 @@ static inline int printw(int c, FILE *fp)
 	return 0;
 }
 
+// an analogy to pileup_func() below
+static int glt3_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
+{
+	pu_data_t *d = (pu_data_t*)data;
+	bam_maqindel_ret_t *r = 0;
+	int rb, *proposed_indels = 0;
+	glf1_t *g;
+	glf3_t *g3;
+
+	if (d->fai == 0) {
+		fprintf(pysamerr, "[glt3_func] reference sequence is required for generating GLT. Abort!\n");
+		exit(1);
+	}
+	if (d->hash) { // only output a list of sites
+		khint_t k = kh_get(64, d->hash, (uint64_t)tid<<32|pos);
+		if (k == kh_end(d->hash)) return 0;
+		proposed_indels = kh_val(d->hash, k);
+	}
+	g3 = glf3_init1();
+	if (d->fai && (int)tid != d->tid) {
+		if (d->ref) { // then write the end mark
+			g3->rtype = GLF3_RTYPE_END;
+			glf3_write1(d->fp_glf, g3);
+		}
+		glf3_ref_write(d->fp_glf, d->h->target_name[tid], d->h->target_len[tid]); // write reference
+		free(d->ref);
+		d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
+		d->tid = tid;
+		d->last_pos = 0;
+	}
+	rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
+	g = bam_maqcns_glfgen(n, pu, bam_nt16_table[rb], d->c);
+	memcpy(g3, g, sizeof(glf1_t));
+	g3->rtype = GLF3_RTYPE_SUB;
+	g3->offset = pos - d->last_pos;
+	d->last_pos = pos;
+	glf3_write1(d->fp_glf, g3);
+    if (pos < d->len) {
+        int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
+		if (proposed_indels)
+			r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+		else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
+	}
+	if (r) { // then write indel line
+		int het = 3 * n, min;
+		min = het;
+		if (min > r->gl[0]) min = r->gl[0];
+		if (min > r->gl[1]) min = r->gl[1];
+		g3->ref_base = 0;
+		g3->rtype = GLF3_RTYPE_INDEL;
+		memset(g3->lk, 0, 10);
+		g3->lk[0] = r->gl[0] - min < 255? r->gl[0] - min : 255;
+		g3->lk[1] = r->gl[1] - min < 255? r->gl[1] - min : 255;
+		g3->lk[2] = het - min < 255? het - min : 255;
+		g3->offset = 0;
+		g3->indel_len[0] = r->indel1;
+		g3->indel_len[1] = r->indel2;
+		g3->min_lk = min < 255? min : 255;
+		g3->max_len = (abs(r->indel1) > abs(r->indel2)? abs(r->indel1) : abs(r->indel2)) + 1;
+		g3->indel_seq[0] = strdup(r->s[0]+1);
+		g3->indel_seq[1] = strdup(r->s[1]+1);
+		glf3_write1(d->fp_glf, g3);
+		bam_maqindel_ret_destroy(r);
+	}
+	free(g);
+	glf3_destroy1(g3);
+	return 0;
+}
+
 static inline void pileup_seq(const bam_pileup1_t *p, int pos, int ref_len, const char *ref)
 {
 	int j;
@@ -60,6 +214,317 @@ static inline void pileup_seq(const bam_pileup1_t *p, int pos, int ref_len, cons
 	if (p->is_tail) putchar('$');
 }
 
+static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
+{
+	pu_data_t *d = (pu_data_t*)data;
+	bam_maqindel_ret_t *r = 0;
+	int i, rb, rms_mapq = -1, *proposed_indels = 0;
+	uint64_t rms_aux;
+	uint32_t cns = 0;
+
+	// if GLF is required, suppress -c completely
+	if (d->format & BAM_PLF_GLF) return glt3_func(tid, pos, n, pu, data);
+	// if d->hash is initialized, only output the sites in the hash table
+	if (d->hash) {
+		khint_t k = kh_get(64, d->hash, (uint64_t)tid<<32|pos);
+		if (k == kh_end(d->hash)) return 0;
+		proposed_indels = kh_val(d->hash, k);
+	}
+	// update d->ref if necessary
+	if (d->fai && (int)tid != d->tid) {
+		free(d->ref);
+		d->ref = faidx_fetch_seq(d->fai, d->h->target_name[tid], 0, 0x7fffffff, &d->len);
+		d->tid = tid;
+	}
+	rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
+	// when the indel-only mode is asked for, return if no reads mapped with indels
+	if (d->format & BAM_PLF_INDEL_ONLY) {
+		for (i = 0; i < n; ++i)
+			if (pu[i].indel != 0) break;
+		if (i == n) return 0;
+	}
+	// call the consensus and indel
+	if (d->format & BAM_PLF_CNS) { // call consensus
+		if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE)) { // use a random base or the 1st base as the consensus call
+			const bam_pileup1_t *p = (d->format & BAM_PLF_1STBASE)? pu : pu + (int)(drand48() * n);
+			int q = bam1_qual(p->b)[p->qpos];
+			int mapQ = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
+			uint32_t b = bam1_seqi(bam1_seq(p->b), p->qpos);
+			cns = b<<28 | 0xf<<24 | mapQ<<16 | q<<8;
+		} else if (d->format & BAM_PLF_ALLBASE) { // collapse all bases
+			uint64_t rmsQ = 0;
+			uint32_t b = 0;
+			for (i = 0; i < n; ++i) {
+				const bam_pileup1_t *p = pu + i;
+				int q = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
+				b |= bam1_seqi(bam1_seq(p->b), p->qpos);
+				rmsQ += q * q;
+			}
+			rmsQ = (uint64_t)(sqrt((double)rmsQ / n) + .499);
+			cns = b<<28 | 0xf<<24 | rmsQ<<16 | 60<<8;
+		} else {
+			glf1_t *g = bam_maqcns_glfgen(n, pu, bam_nt16_table[rb], d->c);
+			cns = g->depth == 0? (0xfu<<28 | 0xf<<24) : glf2cns(g, (int)(d->c->q_r + .499));
+			free(g);
+		}
+	}
+    if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref && pos < d->len) { // call indels
+        int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
+        if (proposed_indels) // the first element gives the size of the array
+            r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+        else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
+	}
+	// when only variant sites are asked for, test if the site is a variant
+	if ((d->format & BAM_PLF_CNS) && (d->format & BAM_PLF_VAR_ONLY)) {
+		if (!(bam_nt16_table[rb] != 15 && cns>>28 != 15 && cns>>28 != bam_nt16_table[rb])) { // not a SNP
+			if (!(r && (r->gt == 2 || strcmp(r->s[r->gt], "*")))) { // not an indel
+				if (r) bam_maqindel_ret_destroy(r);
+				return 0;
+			}
+		}
+	}
+	// print the first 3 columns
+	fputs(d->h->target_name[tid], stdout); putchar('\t');
+	printw(pos+1, stdout); putchar('\t'); putchar(rb); putchar('\t');
+	// print consensus information if required
+	if (d->format & BAM_PLF_CNS) {
+		putchar(bam_nt16_rev_table[cns>>28]); putchar('\t');
+		printw(cns>>8&0xff, stdout); putchar('\t');
+		printw(cns&0xff, stdout); putchar('\t');
+		printw(cns>>16&0xff, stdout); putchar('\t');
+	}
+	// print pileup sequences
+	printw(n, stdout); putchar('\t');
+	for (i = 0; i < n; ++i)
+		pileup_seq(pu + i, pos, d->len, d->ref);
+	// finalize rms_mapq
+	if (d->format & BAM_PLF_CNS) {
+		for (i = rms_aux = 0; i < n; ++i) {
+			const bam_pileup1_t *p = pu + i;
+			int tmp = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
+			rms_aux += tmp * tmp;
+		}
+		rms_aux = (uint64_t)(sqrt((double)rms_aux / n) + .499);
+		if (rms_mapq < 0) rms_mapq = rms_aux;
+	}
+	putchar('\t');
+	// print quality
+	for (i = 0; i < n; ++i) {
+		const bam_pileup1_t *p = pu + i;
+		int c = bam1_qual(p->b)[p->qpos] + 33;
+		if (c > 126) c = 126;
+		putchar(c);
+	}
+	if (d->format & BAM_PLF_2ND) { // print 2nd calls and qualities
+		const unsigned char *q;
+		putchar('\t');
+		for (i = 0; i < n; ++i) {
+			const bam_pileup1_t *p = pu + i;
+			q = bam_aux_get(p->b, "E2");
+			putchar(q? q[p->qpos + 1] : 'N');
+		}
+		putchar('\t');
+		for (i = 0; i < n; ++i) {
+			const bam_pileup1_t *p = pu + i;
+			q = bam_aux_get(p->b, "U2");
+			putchar(q? q[p->qpos + 1] : '!');
+		}
+	}
+	// print mapping quality if -s is flagged on the command line
+	if (d->format & BAM_PLF_SIMPLE) {
+		putchar('\t');
+		for (i = 0; i < n; ++i) {
+			int c = pu[i].b->core.qual + 33;
+			if (c > 126) c = 126;
+			putchar(c);
+		}
+	}
+	// print read position
+	if (d->format & BAM_PLF_READPOS) {
+		putchar('\t');
+		for (i = 0; i < n; ++i) {
+			int x = pu[i].qpos;
+			int l = pu[i].b->core.l_qseq;
+			printw(x < l/2? x+1 : -((l-1)-x+1), stdout); putchar(',');
+		}
+	}
+	putchar('\n');
+	// print the indel line if r has been calculated. This only happens if:
+	// a) -c or -i are flagged, AND b) the reference sequence is available
+	if (r) {
+		printf("%s\t%d\t*\t", d->h->target_name[tid], pos + 1);
+		if (r->gt < 2) printf("%s/%s\t", r->s[r->gt], r->s[r->gt]);
+		else printf("%s/%s\t", r->s[0], r->s[1]);
+		printf("%d\t%d\t", r->q_cns, r->q_ref);
+		printf("%d\t%d\t", rms_mapq, n);
+		printf("%s\t%s\t", r->s[0], r->s[1]);
+		//printf("%d\t%d\t", r->gl[0], r->gl[1]);
+		printf("%d\t%d\t%d\t", r->cnt1, r->cnt2, r->cnt_anti);
+		printf("%d\t%d\n", r->cnt_ref, r->cnt_ambi);
+		bam_maqindel_ret_destroy(r);
+	}
+	return 0;
+}
+
+int bam_pileup(int argc, char *argv[])
+{
+	int c, is_SAM = 0;
+	char *fn_list = 0, *fn_fa = 0, *fn_pos = 0;
+	pu_data_t *d = (pu_data_t*)calloc(1, sizeof(pu_data_t));
+    d->max_depth = 1024; d->tid = -1; d->mask = BAM_DEF_MASK; d->min_baseQ = 13;
+	d->c = bam_maqcns_init();
+	d->c->errmod = BAM_ERRMOD_MAQ2; // change the default model
+	d->ido = bam_maqindel_opt_init();
+	while ((c = getopt(argc, argv, "st:f:cT:N:r:l:d:im:gI:G:vM:S2aR:PAQ:C:B")) >= 0) {
+		switch (c) {
+		case 'Q': d->c->min_baseQ = atoi(optarg); break;
+		case 'C': d->capQ_thres = atoi(optarg); break;
+		case 'B': d->format |= BAM_PLF_NOBAQ; break;
+		case 'a': d->c->errmod = BAM_ERRMOD_SOAP; break;
+		case 'A': d->c->errmod = BAM_ERRMOD_MAQ; break;
+		case 's': d->format |= BAM_PLF_SIMPLE; break;
+		case 't': fn_list = strdup(optarg); break;
+		case 'l': fn_pos = strdup(optarg); break;
+		case 'f': fn_fa = strdup(optarg); break;
+		case 'T': d->c->theta = atof(optarg); break;
+		case 'N': d->c->n_hap = atoi(optarg); break;
+		case 'r': d->c->het_rate = atof(optarg); d->ido->r_snp = d->c->het_rate; break;
+		case 'M': d->c->cap_mapQ = atoi(optarg); break;
+		case 'd': d->max_depth = atoi(optarg); break;
+		case 'c': d->format |= BAM_PLF_CNS; break;
+		case 'i': d->format |= BAM_PLF_INDEL_ONLY; break;
+		case 'v': d->format |= BAM_PLF_VAR_ONLY; break;
+		case 'm': d->mask = strtol(optarg, 0, 0); break;
+		case 'g': d->format |= BAM_PLF_GLF; break;
+		case '2': d->format |= BAM_PLF_2ND; break;
+		case 'P': d->format |= BAM_PLF_READPOS; break;
+		case 'I': d->ido->q_indel = atoi(optarg); break;
+		case 'G': d->ido->r_indel = atof(optarg); break;
+		case 'S': is_SAM = 1; break;
+		case 'R':
+			if (strcmp(optarg, "random") == 0) d->format |= BAM_PLF_RANBASE;
+			else if (strcmp(optarg, "first") == 0) d->format |= BAM_PLF_1STBASE;
+			else if (strcmp(optarg, "all") == 0) d->format |= BAM_PLF_ALLBASE;
+			else fprintf(pysamerr, "[bam_pileup] unrecognized -R\n");
+			break;
+		default: fprintf(pysamerr, "Unrecognizd option '-%c'.\n", c); return 1;
+		}
+	}
+	if (d->c->errmod != BAM_ERRMOD_MAQ2) d->c->theta += 0.02;
+	if (d->c->theta > 1.0) d->c->theta = 1.0;
+	if (fn_list) is_SAM = 1;
+	if (optind == argc) {
+		fprintf(pysamerr, "\n");
+		fprintf(pysamerr, "Usage:  samtools pileup [options] <in.bam>|<in.sam>\n\n");
+		fprintf(pysamerr, "Option: -s        simple (yet incomplete) pileup format\n");
+		fprintf(pysamerr, "        -S        the input is in SAM\n");
+		fprintf(pysamerr, "        -B        disable BAQ computation\n");
+		fprintf(pysamerr, "        -A        use the original MAQ model for SNP calling (DEPRECATED)\n");
+		fprintf(pysamerr, "        -2        output the 2nd best call and quality\n");
+		fprintf(pysamerr, "        -i        only show lines/consensus with indels\n");
+		fprintf(pysamerr, "        -Q INT    min base quality (possibly capped by BAQ) [%d]\n", d->c->min_baseQ);
+		fprintf(pysamerr, "        -C INT    coefficient for adjusting mapQ of poor mappings [%d]\n", d->capQ_thres);
+		fprintf(pysamerr, "        -m INT    filtering reads with bits in INT [0x%x]\n", d->mask);
+		fprintf(pysamerr, "        -M INT    cap mapping quality at INT [%d]\n", d->c->cap_mapQ);
+        fprintf(pysamerr, "        -d INT    limit maximum depth for indels [%d]\n", d->max_depth);
+		fprintf(pysamerr, "        -t FILE   list of reference sequences (force -S)\n");
+		fprintf(pysamerr, "        -l FILE   list of sites at which pileup is output\n");
+		fprintf(pysamerr, "        -f FILE   reference sequence in the FASTA format\n\n");
+		fprintf(pysamerr, "        -c        compute the consensus sequence\n");
+		fprintf(pysamerr, "        -v        print variants only (for -c)\n");
+		fprintf(pysamerr, "        -g        output in the GLFv3 format (DEPRECATED)\n");
+		fprintf(pysamerr, "        -T FLOAT  theta in maq consensus calling model (for -c) [%.4g]\n", d->c->theta);
+		fprintf(pysamerr, "        -N INT    number of haplotypes in the sample (for -c) [%d]\n", d->c->n_hap);
+		fprintf(pysamerr, "        -r FLOAT  prior of a difference between two haplotypes (for -c) [%.4g]\n", d->c->het_rate);
+		fprintf(pysamerr, "        -G FLOAT  prior of an indel between two haplotypes (for -c) [%.4g]\n", d->ido->r_indel);
+		fprintf(pysamerr, "        -I INT    phred prob. of an indel in sequencing/prep. (for -c) [%d]\n", d->ido->q_indel);
+		fprintf(pysamerr, "\n");
+		fprintf(pysamerr, "Warning: Please use the `mpileup' command instead `pileup'. `Pileup' is deprecated!\n\n");
+		free(fn_list); free(fn_fa); free(d);
+		return 1;
+	}
+	if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE|BAM_PLF_ALLBASE)) d->format |= BAM_PLF_CNS;
+	if (fn_fa) d->fai = fai_load(fn_fa);
+	if (d->format & (BAM_PLF_CNS|BAM_PLF_GLF)) bam_maqcns_prepare(d->c); // consensus calling
+	if (d->format & BAM_PLF_GLF) { // for glf output
+		glf3_header_t *h;
+		h = glf3_header_init();
+		d->fp_glf = bgzf_fdopen(fileno(stdout), "w");
+		glf3_header_write(d->fp_glf, h);
+		glf3_header_destroy(h);
+	}
+	if (d->fai == 0 && (d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)))
+		fprintf(pysamerr, "[bam_pileup] indels will not be called when -f is absent.\n");
+	if (fn_fa && is_SAM && fn_list == 0) fn_list = samfaipath(fn_fa);
+
+	{
+		samfile_t *fp;
+		fp = is_SAM? samopen(argv[optind], "r", fn_list) : samopen(argv[optind], "rb", 0);
+		if (fp == 0 || fp->header == 0) {
+			fprintf(pysamerr, "[bam_pileup] fail to read the header: non-exisiting file or wrong format.\n");
+			return 1;
+		}
+		d->h = fp->header;
+		if (fn_pos) d->hash = load_pos(fn_pos, d->h);
+		{ // run pileup
+			extern int bam_prob_realn(bam1_t *b, const char *ref);
+			extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres);
+			bam1_t *b;
+			int ret, tid, pos, n_plp;
+			bam_plp_t iter;
+			const bam_pileup1_t *plp;
+			b = bam_init1();
+			iter = bam_plp_init(0, 0);
+			bam_plp_set_mask(iter, d->mask);
+			while ((ret = samread(fp, b)) >= 0) {
+				int skip = 0;
+				if ((int)b->core.tid < 0) break;
+				// update d->ref if necessary
+				if (d->fai && (int)b->core.tid != d->tid) {
+					free(d->ref);
+					d->ref = faidx_fetch_seq(d->fai, d->h->target_name[b->core.tid], 0, 0x7fffffff, &d->len);
+					d->tid = b->core.tid;
+				}
+				if (d->ref && (d->format&BAM_PLF_CNS) && !(d->format&BAM_PLF_NOBAQ)) bam_prob_realn(b, d->ref);
+				if (d->ref && (d->format&BAM_PLF_CNS) && d->capQ_thres > 10) {
+					int q = bam_cap_mapQ(b, d->ref, d->capQ_thres);
+					if (q < 0) skip = 1;
+					else if (b->core.qual > q) b->core.qual = q;
+				} else if (b->core.flag&BAM_FUNMAP) skip = 1;
+				else if ((d->format&BAM_PLF_CNS) && (b->core.flag&1) && !(b->core.flag&2)) skip = 1;
+				if (skip) continue;
+				bam_plp_push(iter, b);
+				while ((plp = bam_plp_next(iter, &tid, &pos, &n_plp)) != 0)
+					pileup_func(tid, pos, n_plp, plp, d);
+			}
+			bam_plp_push(iter, 0);
+			while ((plp = bam_plp_next(iter, &tid, &pos, &n_plp)) != 0)
+				pileup_func(tid, pos, n_plp, plp, d);
+			bam_plp_destroy(iter);
+			bam_destroy1(b);
+		}
+		samclose(fp); // d->h will be destroyed here
+	}
+
+	// free
+	if (d->format & BAM_PLF_GLF) bgzf_close(d->fp_glf);
+	if (fn_pos) { // free the hash table
+		khint_t k;
+		for (k = kh_begin(d->hash); k < kh_end(d->hash); ++k)
+			if (kh_exist(d->hash, k)) free(kh_val(d->hash, k));
+		kh_destroy(64, d->hash);
+	}
+	free(fn_pos); free(fn_list); free(fn_fa);
+	if (d->fai) fai_destroy(d->fai);
+	bam_maqcns_destroy(d->c);
+	free(d->ido); free(d->ref); free(d);
+	return 0;
+}
+
+/***********
+ * mpileup *
+ ***********/
+
 #include <assert.h>
 #include "bam2bcf.h"
 #include "sample.h"
@@ -73,9 +538,6 @@ static inline void pileup_seq(const bam_pileup1_t *p, int pos, int ref_len, cons
 #define MPLP_NO_INDEL 0x400
 #define MPLP_EXT_BAQ 0x800
 #define MPLP_ILLUMINA13 0x1000
-#define MPLP_IGNORE_RG 0x2000
-#define MPLP_PRINT_POS 0x4000
-#define MPLP_PRINT_MAPQ 0x8000
 
 void *bed_read(const char *fn);
 void bed_destroy(void *_h);
@@ -150,7 +612,7 @@ static int mplp_func(void *data, bam1_t *b)
 }
 
 static void group_smpl(mplp_pileup_t *m, bam_sample_t *sm, kstring_t *buf,
-					   int n, char *const*fn, int *n_plp, const bam_pileup1_t **plp, int ignore_rg)
+					   int n, char *const*fn, int *n_plp, const bam_pileup1_t **plp)
 {
 	int i, j;
 	memset(m->n_plp, 0, m->n * sizeof(int));
@@ -159,7 +621,7 @@ static void group_smpl(mplp_pileup_t *m, bam_sample_t *sm, kstring_t *buf,
 			const bam_pileup1_t *p = plp[i] + j;
 			uint8_t *q;
 			int id = -1;
-			q = ignore_rg? 0 : bam_aux_get(p->b, "RG");
+			q = bam_aux_get(p->b, "RG");
 			if (q) id = bam_smpl_rg2smid(sm, fn[i], (char*)q+1, buf);
 			if (id < 0) id = bam_smpl_rg2smid(sm, fn[i], 0, buf);
 			if (id < 0 || id >= m->n) {
@@ -181,7 +643,7 @@ 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;
+	int i, tid, pos, *n_plp, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid, max_depth, max_indel_depth;
 	const bam_pileup1_t **plp;
 	bam_mplp_t iter;
 	bam_header_t *h = 0;
@@ -214,7 +676,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 		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);
+		bam_smpl_add(sm, fn[i], h_tmp->text);
 		rghash = bcf_call_add_rg(rghash, h_tmp->text, conf->pl_list);
 		if (conf->reg) {
 			int beg, end;
@@ -228,7 +690,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 				fprintf(pysamerr, "[%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;
+			if (i == 0) beg0 = beg, end0 = end;
 			data[i]->iter = bam_iter_query(idx, tid, beg, end);
 			bam_index_destroy(idx);
 		}
@@ -276,11 +738,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 		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;
+	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)
@@ -296,7 +754,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 		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);
+			if (conf->fai) ref = fai_fetch(conf->fai, h->target_name[tid], &ref_len);
 			for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid;
 			ref_tid = tid;
 		}
@@ -304,7 +762,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 			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);
+			group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp);
 			_ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
 			ref16 = bam_nt16_table[_ref0];
 			for (i = 0; i < gplp.n; ++i)
@@ -331,10 +789,8 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 			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 {
+				if (n_plp[i] == 0) printf("*\t*");
+				else {
 					for (j = 0; j < n_plp[i]; ++j)
 						pileup_seq(plp[i] + j, pos, ref_len, ref);
 					putchar('\t');
@@ -344,21 +800,6 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 						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');
@@ -439,7 +880,6 @@ int bam_mpileup(int argc, char *argv[])
     int nfiles = 0, use_orphan = 0;
 	mplp_conf_t mplp;
 	memset(&mplp, 0, sizeof(mplp_conf_t));
-	#define MPLP_PRINT_POS 0x4000
 	mplp.max_mq = 60;
 	mplp.min_baseQ = 13;
 	mplp.capQ_thres = 0;
@@ -447,7 +887,7 @@ int bam_mpileup(int argc, char *argv[])
 	mplp.openQ = 40; mplp.extQ = 20; mplp.tandemQ = 100;
 	mplp.min_frac = 0.002; mplp.min_support = 1;
 	mplp.flag = MPLP_NO_ORPHAN | MPLP_REALN;
-	while ((c = getopt(argc, argv, "Agf:r:l:M:q:Q:uaRC:BDSd:L:b:P:o:e:h:Im:F:EG:6Os")) >= 0) {
+	while ((c = getopt(argc, argv, "Agf:r:l:M:q:Q:uaRC:BDSd:L:b:P:o:e:h:Im:F:EG:6")) >= 0) {
 		switch (c) {
 		case 'f':
 			mplp.fai = fai_load(optarg);
@@ -461,14 +901,12 @@ int bam_mpileup(int argc, char *argv[])
 		case 'u': mplp.flag |= MPLP_NO_COMP | MPLP_GLF; break;
 		case 'a': mplp.flag |= MPLP_NO_ORPHAN | MPLP_REALN; break;
 		case 'B': mplp.flag &= ~MPLP_REALN; break;
+		case 'R': mplp.flag |= MPLP_REALN; break;
 		case 'D': mplp.flag |= MPLP_FMT_DP; break;
 		case 'S': mplp.flag |= MPLP_FMT_SP; break;
 		case 'I': mplp.flag |= MPLP_NO_INDEL; break;
 		case 'E': mplp.flag |= MPLP_EXT_BAQ; break;
 		case '6': mplp.flag |= MPLP_ILLUMINA13; break;
-		case 'R': mplp.flag |= MPLP_IGNORE_RG; break;
-		case 's': mplp.flag |= MPLP_PRINT_MAPQ; break;
-		case 'O': mplp.flag |= MPLP_PRINT_POS; break;
 		case 'C': mplp.capQ_thres = atoi(optarg); break;
 		case 'M': mplp.max_mq = atoi(optarg); break;
 		case 'q': mplp.min_mq = atoi(optarg); break;
@@ -503,7 +941,6 @@ int bam_mpileup(int argc, char *argv[])
 		fprintf(pysamerr, "       -A           count anomalous read pairs\n");
 		fprintf(pysamerr, "       -B           disable BAQ computation\n");
 		fprintf(pysamerr, "       -b FILE      list of input BAM files [null]\n");
-		fprintf(pysamerr, "       -C INT       parameter for adjusting mapQ; 0 to disable [0]\n");
 		fprintf(pysamerr, "       -d INT       max per-BAM depth to avoid excessive memory usage [%d]\n", mplp.max_depth);
 		fprintf(pysamerr, "       -E           extended BAQ for higher sensitivity but lower specificity\n");
 		fprintf(pysamerr, "       -f FILE      faidx indexed reference sequence file [null]\n");
@@ -511,14 +948,11 @@ int bam_mpileup(int argc, char *argv[])
 		fprintf(pysamerr, "       -l FILE      list of positions (chr pos) or regions (BED) [null]\n");
 		fprintf(pysamerr, "       -M INT       cap mapping quality at INT [%d]\n", mplp.max_mq);
 		fprintf(pysamerr, "       -r STR       region in which pileup is generated [null]\n");
-		fprintf(pysamerr, "       -R           ignore RG tags\n");
 		fprintf(pysamerr, "       -q INT       skip alignments with mapQ smaller than INT [%d]\n", mplp.min_mq);
 		fprintf(pysamerr, "       -Q INT       skip bases with baseQ/BAQ smaller than INT [%d]\n", mplp.min_baseQ);
 		fprintf(pysamerr, "\nOutput options:\n\n");
 		fprintf(pysamerr, "       -D           output per-sample DP in BCF (require -g/-u)\n");
 		fprintf(pysamerr, "       -g           generate BCF output (genotype likelihoods)\n");
-		fprintf(pysamerr, "       -O           output base positions on reads (disabled by -g/-u)\n");
-		fprintf(pysamerr, "       -s           output mapping quality (disabled by -g/-u)\n");
 		fprintf(pysamerr, "       -S           output per-sample strand bias P-value in BCF (require -g/-u)\n");
 		fprintf(pysamerr, "       -u           generate uncompress BCF output\n");
 		fprintf(pysamerr, "\nSNP/INDEL genotype likelihoods options (effective with `-g' or `-u'):\n\n");