void bcf_sr_sort_destroy(sr_sort_t *srt)
{
free(srt->active);
if ( srt->var_str2int ) khash_str2int_destroy_free(srt->var_str2int);
if ( srt->grp_str2int ) khash_str2int_destroy_free(srt->grp_str2int);
int i;
for (i=0; i<srt->nsr; i++) free(srt->vcf_buf[i].rec);
free(srt->vcf_buf);
for (i=0; i<srt->mvar; i++)
{
free(srt->var[i].str);
free(srt->var[i].vcf);
free(srt->var[i].rec);
kbs_destroy(srt->var[i].mask);
}
free(srt->var);
for (i=0; i<srt->mgrp; i++)
free(srt->grp[i].var);
free(srt->grp);
for (i=0; i<srt->mvset; i++)
{
kbs_destroy(srt->vset[i].mask);
free(srt->vset[i].var);
}
free(srt->vset);
free(srt->str.s);
free(srt->off);
free(srt->charp);
free(srt->cnt);
free(srt->pmat);
memset(srt,0,sizeof(*srt));
}
void bam_smpl_destroy(bam_smpl_t *bsmpl)
{
if ( !bsmpl ) return;
if ( bsmpl->name2idx ) khash_str2int_destroy_free(bsmpl->name2idx);
if ( bsmpl->sample_list ) khash_str2str_destroy_free_all(bsmpl->sample_list);
if ( bsmpl->rg_list ) khash_str2str_destroy_free_all(bsmpl->rg_list);
int i;
for (i=0; i<bsmpl->nfiles; i++)
{
file_t *file = &bsmpl->files[i];
if ( file->rg2idx ) khash_str2int_destroy_free(file->rg2idx);
free(file->fname);
}
free(bsmpl->smpl);
free(bsmpl->files);
free(bsmpl->tmp.s);
free(bsmpl);
}
static void list_columns(args_t *args)
{
void *has_sample = NULL;
if ( args->sample_list )
{
has_sample = khash_str2int_init();
int i, nsmpl;
char **smpl = hts_readlist(args->sample_list, args->sample_is_file, &nsmpl);
for (i=0; i<nsmpl; i++) khash_str2int_inc(has_sample, smpl[i]);
free(smpl);
}
int i;
bcf_sr_t *reader = &args->files->readers[0];
for (i=0; i<bcf_hdr_nsamples(reader->header); i++)
{
if ( has_sample && !khash_str2int_has_key(has_sample, reader->header->samples[i]) ) continue;
printf("%s\n", reader->header->samples[i]);
}
if ( has_sample )
khash_str2int_destroy_free(has_sample);
}
int bam_mpileup(int argc, char *argv[])
{
int c;
const char *file_list = NULL;
char **fn = NULL;
int nfiles = 0, use_orphan = 0;
mplp_conf_t mplp;
memset(&mplp, 0, sizeof(mplp_conf_t));
mplp.min_baseQ = 13;
mplp.capQ_thres = 0;
mplp.max_depth = 250; mplp.max_indel_depth = 250;
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 | MPLP_SMART_OVERLAPS;
mplp.argc = argc; mplp.argv = argv;
mplp.rflag_filter = BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP;
mplp.output_fname = NULL;
static const struct option lopts[] =
{
{"rf", required_argument, NULL, 1}, // require flag
{"ff", required_argument, NULL, 2}, // filter flag
{"incl-flags", required_argument, NULL, 1},
{"excl-flags", required_argument, NULL, 2},
{"output", required_argument, NULL, 3},
{"open-prob", required_argument, NULL, 4},
{"illumina1.3+", no_argument, NULL, '6'},
{"count-orphans", no_argument, NULL, 'A'},
{"bam-list", required_argument, NULL, 'b'},
{"no-BAQ", no_argument, NULL, 'B'},
{"no-baq", no_argument, NULL, 'B'},
{"adjust-MQ", required_argument, NULL, 'C'},
{"adjust-mq", required_argument, NULL, 'C'},
{"max-depth", required_argument, NULL, 'd'},
{"redo-BAQ", no_argument, NULL, 'E'},
{"redo-baq", no_argument, NULL, 'E'},
{"fasta-ref", required_argument, NULL, 'f'},
{"exclude-RG", required_argument, NULL, 'G'},
{"exclude-rg", required_argument, NULL, 'G'},
{"positions", required_argument, NULL, 'l'},
{"region", required_argument, NULL, 'r'},
{"ignore-RG", no_argument, NULL, 'R'},
{"ignore-rg", no_argument, NULL, 'R'},
{"min-MQ", required_argument, NULL, 'q'},
{"min-mq", required_argument, NULL, 'q'},
{"min-BQ", required_argument, NULL, 'Q'},
{"min-bq", required_argument, NULL, 'Q'},
{"ignore-overlaps", no_argument, NULL, 'x'},
{"BCF", no_argument, NULL, 'g'},
{"bcf", no_argument, NULL, 'g'},
{"VCF", no_argument, NULL, 'v'},
{"vcf", no_argument, NULL, 'v'},
{"output-BP", no_argument, NULL, 'O'},
{"output-bp", no_argument, NULL, 'O'},
{"output-MQ", no_argument, NULL, 's'},
{"output-mq", no_argument, NULL, 's'},
{"output-tags", required_argument, NULL, 't'},
{"uncompressed", no_argument, NULL, 'u'},
{"ext-prob", required_argument, NULL, 'e'},
{"gap-frac", required_argument, NULL, 'F'},
{"tandem-qual", required_argument, NULL, 'h'},
{"skip-indels", no_argument, NULL, 'I'},
{"max-idepth", required_argument, NULL, 'L'},
{"min-ireads ", required_argument, NULL, 'm'},
{"per-sample-mF", no_argument, NULL, 'p'},
{"per-sample-mf", no_argument, NULL, 'p'},
{"platforms", required_argument, NULL, 'P'},
{NULL, 0, NULL, 0}
};
while ((c = getopt_long(argc, argv, "Agf:r:l:q:Q:uRC:BDSd:L:b:P:po:e:h:Im:F:EG:6OsVvxt:",lopts,NULL)) >= 0) {
switch (c) {
case 'x': mplp.flag &= ~MPLP_SMART_OVERLAPS; break;
case 1 :
mplp.rflag_require = bam_str2flag(optarg);
if ( mplp.rflag_require<0 ) { fprintf(stderr,"Could not parse --rf %s\n", optarg); return 1; }
break;
case 2 :
mplp.rflag_filter = bam_str2flag(optarg);
if ( mplp.rflag_filter<0 ) { fprintf(stderr,"Could not parse --ff %s\n", optarg); return 1; }
break;
case 3 : mplp.output_fname = optarg; break;
case 4 : mplp.openQ = atoi(optarg); break;
case 'f':
mplp.fai = fai_load(optarg);
if (mplp.fai == 0) return 1;
mplp.fai_fname = optarg;
break;
case 'd': mplp.max_depth = atoi(optarg); break;
case 'r': mplp.reg = strdup(optarg); break;
case 'l':
// In the original version the whole BAM was streamed which is inefficient
// with few BED intervals and big BAMs. Todo: devise a heuristic to determine
// best strategy, that is streaming or jumping.
mplp.bed = bed_read(optarg);
if (!mplp.bed) { print_error_errno("Could not read file \"%s\"", optarg); return 1; }
break;
case 'P': mplp.pl_list = strdup(optarg); break;
case 'p': mplp.flag |= MPLP_PER_SAMPLE; break;
case 'g': mplp.flag |= MPLP_BCF; break;
case 'v': mplp.flag |= MPLP_BCF | MPLP_VCF; break;
case 'u': mplp.flag |= MPLP_NO_COMP | MPLP_BCF; break;
case 'B': mplp.flag &= ~MPLP_REALN; break;
case 'D': mplp.fmt_flag |= B2B_FMT_DP; fprintf(stderr, "[warning] samtools mpileup option `-D` is functional, but deprecated. Please switch to `-t DP` in future.\n"); break;
case 'S': mplp.fmt_flag |= B2B_FMT_SP; fprintf(stderr, "[warning] samtools mpileup option `-S` is functional, but deprecated. Please switch to `-t SP` in future.\n"); break;
case 'V': mplp.fmt_flag |= B2B_FMT_DV; fprintf(stderr, "[warning] samtools mpileup option `-V` is functional, but deprecated. Please switch to `-t DV` in future.\n"); break;
case 'I': mplp.flag |= MPLP_NO_INDEL; break;
case 'E': mplp.flag |= MPLP_REDO_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 'q': mplp.min_mq = atoi(optarg); break;
case 'Q': mplp.min_baseQ = atoi(optarg); break;
case 'b': file_list = optarg; break;
case 'o': {
char *end;
long value = strtol(optarg, &end, 10);
// Distinguish between -o INT and -o FILE (a bit of a hack!)
if (*end == '\0') mplp.openQ = value;
else mplp.output_fname = optarg;
}
break;
case 'e': mplp.extQ = atoi(optarg); break;
case 'h': mplp.tandemQ = atoi(optarg); break;
case 'A': use_orphan = 1; break;
case 'F': mplp.min_frac = atof(optarg); break;
case 'm': mplp.min_support = atoi(optarg); break;
case 'L': mplp.max_indel_depth = atoi(optarg); break;
case 'G': {
FILE *fp_rg;
char buf[1024];
mplp.rghash = khash_str2int_init();
if ((fp_rg = fopen(optarg, "r")) == 0)
fprintf(stderr, "(%s) Fail to open file %s. Continue anyway.\n", __func__, optarg);
while (!feof(fp_rg) && fscanf(fp_rg, "%s", buf) > 0) // this is not a good style, but forgive me...
khash_str2int_inc(mplp.rghash, strdup(buf));
fclose(fp_rg);
}
break;
case 't': mplp.fmt_flag |= parse_format_flag(optarg); break;
default:
fprintf(stderr,"Invalid option: '%c'\n", c);
return 1;
}
}
if ( !(mplp.flag&MPLP_REALN) && mplp.flag&MPLP_REDO_BAQ )
{
fprintf(stderr,"Error: The -B option cannot be combined with -E\n");
return 1;
}
if (use_orphan) mplp.flag &= ~MPLP_NO_ORPHAN;
if (argc == 1)
{
print_usage(stderr, &mplp);
return 1;
}
int ret;
if (file_list) {
if ( read_file_list(file_list,&nfiles,&fn) ) return 1;
ret = mpileup(&mplp,nfiles,fn);
for (c=0; c<nfiles; c++) free(fn[c]);
free(fn);
}
else
ret = mpileup(&mplp, argc - optind, argv + optind);
if (mplp.rghash) khash_str2int_destroy_free(mplp.rghash);
free(mplp.reg); free(mplp.pl_list);
if (mplp.fai) fai_destroy(mplp.fai);
if (mplp.bed) bed_destroy(mplp.bed);
return ret;
}
/*
The logic of this function is a bit complicated because we want to work
also with broken bams containing read groups that are not listed in the
header. The desired behavior is as follows:
- when -G is given, read groups which are not listed in the header must
be given explicitly using the "?" symbol in -G.
Otherwise:
- if the bam has no header, all reads in the file are assigned to a
single sample named after the file
- if there is at least one sample defined in the header, reads with no
read group id or with a read group id not listed in the header are
assigned to the first sample encountered in the header
*/
int bam_smpl_add_bam(bam_smpl_t *bsmpl, char *bam_hdr, const char *fname)
{
bsmpl->nfiles++;
bsmpl->files = (file_t*) realloc(bsmpl->files,bsmpl->nfiles*sizeof(file_t));
file_t *file = &bsmpl->files[bsmpl->nfiles-1];
memset(file,0,sizeof(file_t));
file->fname = strdup(fname);
file->default_idx = -1;
if ( bsmpl->ignore_rg || !bam_hdr )
{
// The option --ignore-RG is set or there is no BAM header: use the file name as the sample name
bsmpl_add_readgroup(bsmpl,file,"*",file->fname);
return bsmpl->nfiles-1;
}
void *bam_smpls = khash_str2int_init();
int first_smpl = -1, nskipped = 0;
const char *p = bam_hdr, *q, *r;
while (p != NULL && (q = strstr(p, "@RG")) != 0)
{
char *eol = strchr(q + 3, '\n');
if (q > bam_hdr && *(q - 1) != '\n') { // @RG must be at start of line
p = eol;
continue;
}
p = q + 3;
if ((q = strstr(p, "\tID:")) != 0) q += 4;
if ((r = strstr(p, "\tSM:")) != 0) r += 4;
if (r && q)
{
char *u, *v;
int ioq, ior;
for (u = (char*)q; *u && *u != '\t' && *u != '\n'; ++u);
for (v = (char*)r; *v && *v != '\t' && *v != '\n'; ++v);
ioq = *u; ior = *v; *u = *v = '\0';
// q now points to a null terminated read group id
// r points to a null terminated sample name
if ( !strcmp("*",q) || !strcmp("?",q) )
error("Error: the read group IDs \"*\" and \"?\" have a special meaning in the mpileup code. Please fix the code or the bam: %s\n", fname);
int accept_rg = 1;
if ( bsmpl->sample_list )
{
// restrict samples based on the -s/-S options
char *name = khash_str2str_get(bsmpl->sample_list,r);
if ( bsmpl->sample_logic==0 )
accept_rg = name ? 0 : 1;
else if ( !name )
accept_rg = 0;
else
r = name;
}
if ( accept_rg && bsmpl->rg_list )
{
// restrict readgroups based on the -G option, possibly renaming the sample
accept_rg = bsmpl_keep_readgroup(bsmpl,file,q,&r);
}
if ( accept_rg )
bsmpl_add_readgroup(bsmpl,file,q,r);
else
{
bsmpl_add_readgroup(bsmpl,file,q,NULL); // ignore this RG but note that it was seen in the header
nskipped++;
}
if ( first_smpl<0 )
khash_str2int_get(bsmpl->name2idx,r,&first_smpl);
if ( !khash_str2int_has_key(bam_smpls,r) )
khash_str2int_inc(bam_smpls,strdup(r));
*u = ioq; *v = ior;
}
else
break;
p = eol;
}
int nsmpls = khash_str2int_size(bam_smpls);
khash_str2int_destroy_free(bam_smpls);
const char *smpl_name = NULL;
int accept_null_rg = 1;
if ( bsmpl->rg_list && !bsmpl_keep_readgroup(bsmpl,file,"?",&smpl_name) ) accept_null_rg = 0;
if ( bsmpl->sample_list && first_smpl==-1 ) accept_null_rg = 0;
if ( !accept_null_rg && first_smpl==-1 )
{
// no suitable read group is available in this bam: ignore the whole file.
free(file->fname);
if ( file->rg2idx ) khash_str2int_destroy_free(file->rg2idx);
bsmpl->nfiles--;
return -1;
}
if ( !accept_null_rg ) return bsmpl->nfiles-1;
if ( nsmpls==1 && !nskipped )
{
file->default_idx = first_smpl;
return bsmpl->nfiles-1;
}
if ( !smpl_name ) smpl_name = first_smpl==-1 ? file->fname : bsmpl->smpl[first_smpl];
bsmpl_add_readgroup(bsmpl,file,"?",smpl_name);
return bsmpl->nfiles-1;
}
