static bcf_hdr_t *strip_header(bcf_hdr_t *src, bcf_hdr_t *dst)
{
bcf_hrec_t *src_hrec, *dst_hrec, *tmp;
bcf_hdr_t *out = bcf_hdr_init("r");
int i;
for (i=0; i<dst->nhrec; i++)
{
// first insert lines which do not code BCF ids, their order does not matter
dst_hrec = dst->hrec[i];
if ( dst_hrec->type==BCF_HL_FLT || dst_hrec->type==BCF_HL_INFO || dst_hrec->type==BCF_HL_FMT || dst_hrec->type== BCF_HL_CTG ) continue;
bcf_hdr_add_hrec(out, bcf_hrec_dup(dst_hrec));
}
for (i=0; i<src->nhrec; i++)
{
// now transfer header lines which define BCF ids
src_hrec = src->hrec[i];
if ( src_hrec->type==BCF_HL_FLT || src_hrec->type==BCF_HL_INFO || src_hrec->type==BCF_HL_FMT || src_hrec->type== BCF_HL_CTG )
{
int j = bcf_hrec_find_key(src_hrec, "ID");
dst_hrec = bcf_hdr_get_hrec(dst, src_hrec->type, "ID", src_hrec->vals[j], NULL);
if ( !dst_hrec ) continue;
tmp = bcf_hrec_dup(dst_hrec);
j = bcf_hrec_find_key(src_hrec, "IDX");
if ( j>=0 )
{
j = atoi(src_hrec->vals[j]);
hrec_add_idx(tmp, j);
}
bcf_hdr_add_hrec(out, tmp);
}
}
bcf_hdr_sync(out);
for (i=0; i<dst->nhrec; i++)
{
// finally add new structured fields
dst_hrec = dst->hrec[i];
if ( dst_hrec->type==BCF_HL_FLT || dst_hrec->type==BCF_HL_INFO || dst_hrec->type==BCF_HL_FMT || dst_hrec->type== BCF_HL_CTG )
{
int j = bcf_hrec_find_key(dst_hrec, "ID");
tmp = bcf_hdr_get_hrec(out, dst_hrec->type, "ID", dst_hrec->vals[j], NULL);
if ( !tmp )
bcf_hdr_add_hrec(out, bcf_hrec_dup(dst_hrec));
}
}
for (i=0; i<dst->n[BCF_DT_SAMPLE]; i++) bcf_hdr_add_sample(out, dst->samples[i]);
bcf_hdr_add_sample(out, NULL);
bcf_hdr_destroy(dst);
return out;
}
void bcf_hdr_merge(bcf_hdr_t *hw, const bcf_hdr_t *_hr, const char *clash_prefix)
{
bcf_hdr_t *hr = (bcf_hdr_t*)_hr;
// header lines
int i, nw_ori = hw->nhrec;
for (i=0; i<hr->nhrec; i++)
{
if ( hr->hrec[i]->type==BCF_HL_GEN && hr->hrec[i]->value )
{
int j;
for (j=0; j<nw_ori; j++)
{
if ( hw->hrec[j]->type!=BCF_HL_GEN ) continue;
if ( !strcmp(hr->hrec[i]->key,hw->hrec[j]->key) && !strcmp(hr->hrec[i]->value,hw->hrec[j]->value) ) break;
}
if ( j>=nw_ori )
bcf_hdr_add_hrec(hw, bcf_hrec_dup(hr->hrec[i]));
}
else
{
bcf_hrec_t *rec = bcf_hdr_get_hrec(hw, hr->hrec[i]->type, hr->hrec[i]->vals[0]);
if ( !rec )
bcf_hdr_add_hrec(hw, bcf_hrec_dup(hr->hrec[i]));
}
}
// samples
for (i=0; i<bcf_hdr_nsamples(hr); i++)
{
char *name = hr->samples[i];
if ( bcf_hdr_id2int(hw, BCF_DT_SAMPLE, name)!=-1 )
{
// there is a sample with the same name
int len = strlen(hr->samples[i]) + strlen(clash_prefix) + 1;
name = (char*) malloc(sizeof(char)*(len+1));
sprintf(name,"%s:%s",clash_prefix,hr->samples[i]);
bcf_hdr_add_sample(hw,name);
free(name);
}
else
bcf_hdr_add_sample(hw,name);
}
}
void bcf_hdr_parse_sample_line(bcf_hdr_t *h, const char *str)
{
int i = 0;
const char *p, *q;
vdict_t *d = (vdict_t*)h->dict[BCF_DT_ID];
// add samples
d = (vdict_t*)h->dict[BCF_DT_SAMPLE];
for (p = q = str;; ++q) {
if (*q != '\t' && *q != 0) continue;
if (++i > 9) {
char *s;
s = (char*)malloc(q - p + 1);
strncpy(s, p, q - p);
s[q - p] = 0;
bcf_hdr_add_sample(h,s);
}
if (*q == 0) break;
p = q + 1;
}
}
static int mpileup(mplp_conf_t *conf)
{
if (conf->nfiles == 0) {
fprintf(stderr,"[%s] no input file/data given\n", __func__);
exit(EXIT_FAILURE);
}
mplp_ref_t mp_ref = MPLP_REF_INIT;
conf->gplp = (mplp_pileup_t *) calloc(1,sizeof(mplp_pileup_t));
conf->mplp_data = (mplp_aux_t**) calloc(conf->nfiles, sizeof(mplp_aux_t*));
conf->plp = (const bam_pileup1_t**) calloc(conf->nfiles, sizeof(bam_pileup1_t*));
conf->n_plp = (int*) calloc(conf->nfiles, sizeof(int));
// Allow to run mpileup on multiple regions in one go. This comes at cost: the bai index
// must be kept in the memory for the whole time which can be a problem with many bams.
// Therefore if none or only one region is requested, we initialize the bam iterator as
// before and free the index. Only when multiple regions are queried, we keep the index.
int nregs = 0;
if ( conf->reg_fname )
{
if ( conf->reg_is_file )
{
conf->reg = regidx_init(conf->reg_fname,NULL,NULL,0,NULL);
if ( !conf->reg ) {
fprintf(stderr,"Could not parse the regions: %s\n", conf->reg_fname);
exit(EXIT_FAILURE);
}
}
else
{
conf->reg = regidx_init(NULL,regidx_parse_reg,NULL,sizeof(char*),NULL);
if ( regidx_insert_list(conf->reg,conf->reg_fname,',') !=0 ) {
fprintf(stderr,"Could not parse the regions: %s\n", conf->reg_fname);
exit(EXIT_FAILURE);
}
}
nregs = regidx_nregs(conf->reg);
conf->reg_itr = regitr_init(conf->reg);
regitr_loop(conf->reg_itr); // region iterator now positioned at the first region
}
// read the header of each file in the list and initialize data
// beware: mpileup has always assumed that tid's are consistent in the headers, add sanity check at least!
bam_hdr_t *hdr = NULL; // header of first file in input list
int i;
for (i = 0; i < conf->nfiles; ++i) {
bam_hdr_t *h_tmp;
conf->mplp_data[i] = (mplp_aux_t*) calloc(1, sizeof(mplp_aux_t));
conf->mplp_data[i]->fp = sam_open(conf->files[i], "rb");
if ( !conf->mplp_data[i]->fp )
{
fprintf(stderr, "[%s] failed to open %s: %s\n", __func__, conf->files[i], strerror(errno));
exit(EXIT_FAILURE);
}
if (hts_set_opt(conf->mplp_data[i]->fp, CRAM_OPT_DECODE_MD, 0)) {
fprintf(stderr, "Failed to set CRAM_OPT_DECODE_MD value\n");
exit(EXIT_FAILURE);
}
if (conf->fai_fname && hts_set_fai_filename(conf->mplp_data[i]->fp, conf->fai_fname) != 0) {
fprintf(stderr, "[%s] failed to process %s: %s\n",
__func__, conf->fai_fname, strerror(errno));
exit(EXIT_FAILURE);
}
conf->mplp_data[i]->conf = conf;
conf->mplp_data[i]->ref = &mp_ref;
h_tmp = sam_hdr_read(conf->mplp_data[i]->fp);
if ( !h_tmp ) {
fprintf(stderr,"[%s] fail to read the header of %s\n", __func__, conf->files[i]);
exit(EXIT_FAILURE);
}
conf->mplp_data[i]->h = i ? hdr : h_tmp; // for j==0, "h" has not been set yet
conf->mplp_data[i]->bam_id = bam_smpl_add_bam(conf->bsmpl,h_tmp->text,conf->files[i]);
if ( conf->mplp_data[i]->bam_id<0 )
{
// no usable readgroups in this bam, it can be skipped
sam_close(conf->mplp_data[i]->fp);
free(conf->mplp_data[i]);
bam_hdr_destroy(h_tmp);
free(conf->files[i]);
if ( i+1<conf->nfiles ) memmove(&conf->files[i],&conf->files[i+1],sizeof(*conf->files)*(conf->nfiles-i-1));
conf->nfiles--;
i--;
continue;
}
if (conf->reg) {
hts_idx_t *idx = sam_index_load(conf->mplp_data[i]->fp, conf->files[i]);
if (idx == NULL) {
fprintf(stderr, "[%s] fail to load index for %s\n", __func__, conf->files[i]);
exit(EXIT_FAILURE);
}
conf->buf.l = 0;
ksprintf(&conf->buf,"%s:%u-%u",conf->reg_itr->seq,conf->reg_itr->beg+1,conf->reg_itr->end+1);
conf->mplp_data[i]->iter = sam_itr_querys(idx, conf->mplp_data[i]->h, conf->buf.s);
if ( !conf->mplp_data[i]->iter )
{
conf->mplp_data[i]->iter = sam_itr_querys(idx, conf->mplp_data[i]->h, conf->reg_itr->seq);
if ( conf->mplp_data[i]->iter ) {
fprintf(stderr,"[E::%s] fail to parse region '%s'\n", __func__, conf->buf.s);
exit(EXIT_FAILURE);
}
fprintf(stderr,"[E::%s] the sequence \"%s\" not found: %s\n",__func__,conf->reg_itr->seq,conf->files[i]);
exit(EXIT_FAILURE);
}
if ( nregs==1 ) // no need to keep the index in memory
hts_idx_destroy(idx);
else
conf->mplp_data[i]->idx = idx;
}
if ( !hdr ) hdr = h_tmp; /* save the header of first file in list */
else {
// FIXME: check consistency between h and h_tmp
bam_hdr_destroy(h_tmp);
// we store only the first file's header; it's (alleged to be)
// compatible with the i-th file's target_name lookup needs
conf->mplp_data[i]->h = hdr;
}
}
// allocate data storage proportionate to number of samples being studied sm->n
bam_smpl_get_samples(conf->bsmpl, &conf->gplp->n);
conf->gplp->n_plp = (int*) calloc(conf->gplp->n, sizeof(int));
conf->gplp->m_plp = (int*) calloc(conf->gplp->n, sizeof(int));
conf->gplp->plp = (bam_pileup1_t**) calloc(conf->gplp->n, sizeof(bam_pileup1_t*));
fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, conf->gplp->n, conf->nfiles);
// write the VCF header
conf->bcf_fp = hts_open(conf->output_fname?conf->output_fname:"-", hts_bcf_wmode(conf->output_type));
if (conf->bcf_fp == NULL) {
fprintf(stderr, "[%s] failed to write to %s: %s\n", __func__, conf->output_fname? conf->output_fname : "standard output", strerror(errno));
exit(EXIT_FAILURE);
}
if ( conf->n_threads ) hts_set_threads(conf->bcf_fp, conf->n_threads);
// BCF header creation
conf->bcf_hdr = bcf_hdr_init("w");
conf->buf.l = 0;
if (conf->record_cmd_line)
{
ksprintf(&conf->buf, "##bcftoolsVersion=%s+htslib-%s\n",bcftools_version(),hts_version());
bcf_hdr_append(conf->bcf_hdr, conf->buf.s);
conf->buf.l = 0;
ksprintf(&conf->buf, "##bcftoolsCommand=mpileup");
for (i=1; i<conf->argc; i++) ksprintf(&conf->buf, " %s", conf->argv[i]);
kputc('\n', &conf->buf);
bcf_hdr_append(conf->bcf_hdr, conf->buf.s);
}
if (conf->fai_fname)
{
conf->buf.l = 0;
ksprintf(&conf->buf, "##reference=file://%s\n", conf->fai_fname);
bcf_hdr_append(conf->bcf_hdr, conf->buf.s);
}
// Translate BAM @SQ tags to BCF ##contig tags
// todo: use/write new BAM header manipulation routines, fill also UR, M5
for (i=0; i<hdr->n_targets; i++)
{
conf->buf.l = 0;
ksprintf(&conf->buf, "##contig=<ID=%s,length=%d>", hdr->target_name[i], hdr->target_len[i]);
bcf_hdr_append(conf->bcf_hdr, conf->buf.s);
}
conf->buf.l = 0;
bcf_hdr_append(conf->bcf_hdr,"##ALT=<ID=*,Description=\"Represents allele(s) other than observed.\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=IDV,Number=1,Type=Integer,Description=\"Maximum number of reads supporting an indel\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=IMF,Number=1,Type=Float,Description=\"Maximum fraction of reads supporting an indel\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw read depth\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=VDB,Number=1,Type=Float,Description=\"Variant Distance Bias for filtering splice-site artefacts in RNA-seq data (bigger is better)\",Version=\"3\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=RPB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=BQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQSB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias (bigger is better)\">");
#if CDF_MWU_TESTS
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=RPB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias [CDF] (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=BQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQSB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias [CDF] (bigger is better)\">");
#endif
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=SGB,Number=1,Type=Float,Description=\"Segregation based metric.\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=MQ0F,Number=1,Type=Float,Description=\"Fraction of MQ0 reads (smaller is better)\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=I16,Number=16,Type=Float,Description=\"Auxiliary tag used for calling, see description of bcf_callret1_t in bam2bcf.h\">");
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=QS,Number=R,Type=Float,Description=\"Auxiliary tag used for calling\">");
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=PL,Number=G,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods\">");
if ( conf->fmt_flag&B2B_FMT_DP )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"Number of high-quality bases\">");
if ( conf->fmt_flag&B2B_FMT_DV )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=DV,Number=1,Type=Integer,Description=\"Number of high-quality non-reference bases\">");
if ( conf->fmt_flag&B2B_FMT_DPR )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
if ( conf->fmt_flag&B2B_INFO_DPR )
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
if ( conf->fmt_flag&B2B_FMT_DP4 )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=DP4,Number=4,Type=Integer,Description=\"Number of high-quality ref-fwd, ref-reverse, alt-fwd and alt-reverse bases\">");
if ( conf->fmt_flag&B2B_FMT_SP )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">");
if ( conf->fmt_flag&B2B_FMT_AD )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=AD,Number=R,Type=Integer,Description=\"Allelic depths\">");
if ( conf->fmt_flag&B2B_FMT_ADF )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=ADF,Number=R,Type=Integer,Description=\"Allelic depths on the forward strand\">");
if ( conf->fmt_flag&B2B_FMT_ADR )
bcf_hdr_append(conf->bcf_hdr,"##FORMAT=<ID=ADR,Number=R,Type=Integer,Description=\"Allelic depths on the reverse strand\">");
if ( conf->fmt_flag&B2B_INFO_AD )
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=AD,Number=R,Type=Integer,Description=\"Total allelic depths\">");
if ( conf->fmt_flag&B2B_INFO_ADF )
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=ADF,Number=R,Type=Integer,Description=\"Total allelic depths on the forward strand\">");
if ( conf->fmt_flag&B2B_INFO_ADR )
bcf_hdr_append(conf->bcf_hdr,"##INFO=<ID=ADR,Number=R,Type=Integer,Description=\"Total allelic depths on the reverse strand\">");
if ( conf->gvcf )
gvcf_update_header(conf->gvcf, conf->bcf_hdr);
int nsmpl;
const char **smpl = bam_smpl_get_samples(conf->bsmpl, &nsmpl);
for (i=0; i<nsmpl; i++)
bcf_hdr_add_sample(conf->bcf_hdr, smpl[i]);
bcf_hdr_write(conf->bcf_fp, conf->bcf_hdr);
conf->bca = bcf_call_init(-1., conf->min_baseQ);
conf->bcr = (bcf_callret1_t*) calloc(nsmpl, sizeof(bcf_callret1_t));
conf->bca->openQ = conf->openQ, conf->bca->extQ = conf->extQ, conf->bca->tandemQ = conf->tandemQ;
conf->bca->min_frac = conf->min_frac;
conf->bca->min_support = conf->min_support;
conf->bca->per_sample_flt = conf->flag & MPLP_PER_SAMPLE;
conf->bc.bcf_hdr = conf->bcf_hdr;
conf->bc.n = nsmpl;
conf->bc.PL = (int32_t*) malloc(15 * nsmpl * sizeof(*conf->bc.PL));
if (conf->fmt_flag)
{
assert( sizeof(float)==sizeof(int32_t) );
conf->bc.DP4 = (int32_t*) malloc(nsmpl * sizeof(int32_t) * 4);
conf->bc.fmt_arr = (uint8_t*) malloc(nsmpl * sizeof(float)); // all fmt_flag fields, float and int32
if ( conf->fmt_flag&(B2B_INFO_DPR|B2B_FMT_DPR|B2B_INFO_AD|B2B_INFO_ADF|B2B_INFO_ADR|B2B_FMT_AD|B2B_FMT_ADF|B2B_FMT_ADR) )
{
// first B2B_MAX_ALLELES fields for total numbers, the rest per-sample
conf->bc.ADR = (int32_t*) malloc((nsmpl+1)*B2B_MAX_ALLELES*sizeof(int32_t));
conf->bc.ADF = (int32_t*) malloc((nsmpl+1)*B2B_MAX_ALLELES*sizeof(int32_t));
for (i=0; i<nsmpl; i++)
{
conf->bcr[i].ADR = conf->bc.ADR + (i+1)*B2B_MAX_ALLELES;
conf->bcr[i].ADF = conf->bc.ADF + (i+1)*B2B_MAX_ALLELES;
}
}
}
// init mpileup
conf->iter = bam_mplp_init(conf->nfiles, mplp_func, (void**)conf->mplp_data);
if ( conf->flag & MPLP_SMART_OVERLAPS ) bam_mplp_init_overlaps(conf->iter);
if ( (double)conf->max_depth * conf->nfiles > 1<<20)
fprintf(stderr, "Warning: Potential memory hog, up to %.0fM reads in the pileup!\n", (double)conf->max_depth*conf->nfiles);
if ( (double)conf->max_depth * conf->nfiles / nsmpl < 250 )
fprintf(stderr, "Note: The maximum per-sample depth with -d %d is %.1fx\n", conf->max_depth,(double)conf->max_depth * conf->nfiles / nsmpl);
bam_mplp_set_maxcnt(conf->iter, conf->max_depth);
conf->max_indel_depth = conf->max_indel_depth * nsmpl;
conf->bcf_rec = bcf_init1();
bam_mplp_constructor(conf->iter, pileup_constructor);
// Run mpileup for multiple regions
if ( nregs )
{
int ireg = 0;
do
{
// first region is already positioned
if ( ireg++ > 0 )
{
conf->buf.l = 0;
ksprintf(&conf->buf,"%s:%u-%u",conf->reg_itr->seq,conf->reg_itr->beg,conf->reg_itr->end);
for (i=0; i<conf->nfiles; i++)
{
hts_itr_destroy(conf->mplp_data[i]->iter);
conf->mplp_data[i]->iter = sam_itr_querys(conf->mplp_data[i]->idx, conf->mplp_data[i]->h, conf->buf.s);
if ( !conf->mplp_data[i]->iter )
{
conf->mplp_data[i]->iter = sam_itr_querys(conf->mplp_data[i]->idx, conf->mplp_data[i]->h, conf->reg_itr->seq);
if ( conf->mplp_data[i]->iter ) {
fprintf(stderr,"[E::%s] fail to parse region '%s'\n", __func__, conf->buf.s);
exit(EXIT_FAILURE);
}
fprintf(stderr,"[E::%s] the sequence \"%s\" not found: %s\n",__func__,conf->reg_itr->seq,conf->files[i]);
exit(EXIT_FAILURE);
}
bam_mplp_reset(conf->iter);
}
}
mpileup_reg(conf,conf->reg_itr->beg,conf->reg_itr->end);
}
while ( regitr_loop(conf->reg_itr) );
}
else
mpileup_reg(conf,0,0);
flush_bcf_records(conf, conf->bcf_fp, conf->bcf_hdr, NULL);
// clean up
free(conf->bc.tmp.s);
bcf_destroy1(conf->bcf_rec);
if (conf->bcf_fp)
{
hts_close(conf->bcf_fp);
bcf_hdr_destroy(conf->bcf_hdr);
bcf_call_destroy(conf->bca);
free(conf->bc.PL);
free(conf->bc.DP4);
free(conf->bc.ADR);
free(conf->bc.ADF);
free(conf->bc.fmt_arr);
free(conf->bcr);
}
if ( conf->gvcf ) gvcf_destroy(conf->gvcf);
free(conf->buf.s);
for (i = 0; i < conf->gplp->n; ++i) free(conf->gplp->plp[i]);
free(conf->gplp->plp); free(conf->gplp->n_plp); free(conf->gplp->m_plp); free(conf->gplp);
bam_mplp_destroy(conf->iter);
bam_hdr_destroy(hdr);
for (i = 0; i < conf->nfiles; ++i) {
if ( nregs>1 ) hts_idx_destroy(conf->mplp_data[i]->idx);
sam_close(conf->mplp_data[i]->fp);
if ( conf->mplp_data[i]->iter) hts_itr_destroy(conf->mplp_data[i]->iter);
free(conf->mplp_data[i]);
}
if ( conf->reg_itr ) regitr_destroy(conf->reg_itr);
free(conf->mplp_data); free(conf->plp); free(conf->n_plp);
free(mp_ref.ref[0]);
free(mp_ref.ref[1]);
return 0;
}
/*
* Performs pileup
* @param conf configuration for this pileup
* @param n number of files specified in fn
* @param fn filenames
*/
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_hdr_t *h = NULL; /* header of first file in input list */
char *ref;
void *rghash = NULL;
FILE *pileup_fp = NULL;
bcf_callaux_t *bca = NULL;
bcf_callret1_t *bcr = NULL;
bcf_call_t bc;
htsFile *bcf_fp = NULL;
bcf_hdr_t *bcf_hdr = NULL;
bam_sample_t *sm = NULL;
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(mplp_aux_t*));
plp = calloc(n, sizeof(bam_pileup1_t*));
n_plp = calloc(n, sizeof(int));
sm = bam_smpl_init();
if (n == 0) {
fprintf(stderr,"[%s] no input file/data given\n", __func__);
exit(1);
}
// read the header of each file in the list and initialize data
for (i = 0; i < n; ++i) {
bam_hdr_t *h_tmp;
data[i] = calloc(1, sizeof(mplp_aux_t));
data[i]->fp = sam_open(fn[i], "rb");
if ( !data[i]->fp )
{
fprintf(stderr, "[%s] failed to open %s: %s\n", __func__, fn[i], strerror(errno));
exit(1);
}
hts_set_fai_filename(data[i]->fp, conf->fai_fname);
data[i]->conf = conf;
h_tmp = sam_hdr_read(data[i]->fp);
if ( !h_tmp ) {
fprintf(stderr,"[%s] fail to read the header of %s\n", __func__, fn[i]);
exit(1);
}
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);
// Collect read group IDs with PL (platform) listed in pl_list (note: fragile, strstr search)
rghash = bcf_call_add_rg(rghash, h_tmp->text, conf->pl_list);
if (conf->reg) {
hts_idx_t *idx = sam_index_load(data[i]->fp, fn[i]);
if (idx == 0) {
fprintf(stderr, "[%s] fail to load index for %s\n", __func__, fn[i]);
exit(1);
}
if ( (data[i]->iter=sam_itr_querys(idx, data[i]->h, conf->reg)) == 0) {
fprintf(stderr, "[E::%s] fail to parse region '%s'\n", __func__, conf->reg);
exit(1);
}
if (i == 0) tid0 = data[i]->iter->tid, beg0 = data[i]->iter->beg, end0 = data[i]->iter->end;
hts_idx_destroy(idx);
}
if (i == 0) h = h_tmp; /* save the header of first file in list */
else {
// FIXME: to check consistency
bam_hdr_destroy(h_tmp);
}
}
// allocate data storage proportionate to number of samples being studied sm->n
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(bam_pileup1_t*));
fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, sm->n, n);
// write the VCF header
if (conf->flag & MPLP_BCF)
{
const char *mode;
if ( conf->flag & MPLP_VCF )
mode = (conf->flag&MPLP_NO_COMP)? "wu" : "wz"; // uncompressed VCF or compressed VCF
else
mode = (conf->flag&MPLP_NO_COMP)? "wub" : "wb"; // uncompressed BCF or compressed BCF
bcf_fp = bcf_open(conf->output_fname? conf->output_fname : "-", mode);
if (bcf_fp == NULL) {
fprintf(stderr, "[%s] failed to write to %s: %s\n", __func__, conf->output_fname? conf->output_fname : "standard output", strerror(errno));
exit(1);
}
bcf_hdr = bcf_hdr_init("w");
kstring_t str = {0,0,0};
ksprintf(&str, "##samtoolsVersion=%s+htslib-%s\n",samtools_version(),hts_version());
bcf_hdr_append(bcf_hdr, str.s);
str.l = 0;
ksprintf(&str, "##samtoolsCommand=samtools mpileup");
for (i=1; i<conf->argc; i++) ksprintf(&str, " %s", conf->argv[i]);
kputc('\n', &str);
bcf_hdr_append(bcf_hdr, str.s);
if (conf->fai_fname)
{
str.l = 0;
ksprintf(&str, "##reference=file://%s\n", conf->fai_fname);
bcf_hdr_append(bcf_hdr, str.s);
}
// todo: use/write new BAM header manipulation routines, fill also UR, M5
for (i=0; i<h->n_targets; i++)
{
str.l = 0;
ksprintf(&str, "##contig=<ID=%s,length=%d>", h->target_name[i], h->target_len[i]);
bcf_hdr_append(bcf_hdr, str.s);
}
free(str.s);
bcf_hdr_append(bcf_hdr,"##ALT=<ID=X,Description=\"Represents allele(s) other than observed.\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=IDV,Number=1,Type=Integer,Description=\"Maximum number of reads supporting an indel\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=IMF,Number=1,Type=Float,Description=\"Maximum fraction of reads supporting an indel\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw read depth\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=VDB,Number=1,Type=Float,Description=\"Variant Distance Bias for filtering splice-site artefacts in RNA-seq data (bigger is better)\",Version=\"3\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=RPB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=BQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQSB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias (bigger is better)\">");
#if CDF_MWU_TESTS
bcf_hdr_append(bcf_hdr,"##INFO=<ID=RPB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias [CDF] (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=BQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQSB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias [CDF] (bigger is better)\">");
#endif
bcf_hdr_append(bcf_hdr,"##INFO=<ID=SGB,Number=1,Type=Float,Description=\"Segregation based metric.\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=MQ0F,Number=1,Type=Float,Description=\"Fraction of MQ0 reads (smaller is better)\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=I16,Number=16,Type=Float,Description=\"Auxiliary tag used for calling, see description of bcf_callret1_t in bam2bcf.h\">");
bcf_hdr_append(bcf_hdr,"##INFO=<ID=QS,Number=R,Type=Float,Description=\"Auxiliary tag used for calling\">");
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=PL,Number=G,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods\">");
if ( conf->fmt_flag&B2B_FMT_DP )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"Number of high-quality bases\">");
if ( conf->fmt_flag&B2B_FMT_DV )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=DV,Number=1,Type=Integer,Description=\"Number of high-quality non-reference bases\">");
if ( conf->fmt_flag&B2B_FMT_DPR )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
if ( conf->fmt_flag&B2B_INFO_DPR )
bcf_hdr_append(bcf_hdr,"##INFO=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
if ( conf->fmt_flag&B2B_FMT_DP4 )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=DP4,Number=4,Type=Integer,Description=\"Number of high-quality ref-fwd, ref-reverse, alt-fwd and alt-reverse bases\">");
if ( conf->fmt_flag&B2B_FMT_SP )
bcf_hdr_append(bcf_hdr,"##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">");
for (i=0; i<sm->n; i++)
bcf_hdr_add_sample(bcf_hdr, sm->smpl[i]);
bcf_hdr_add_sample(bcf_hdr, NULL);
bcf_hdr_write(bcf_fp, bcf_hdr);
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;
bca->per_sample_flt = conf->flag & MPLP_PER_SAMPLE;
bc.bcf_hdr = bcf_hdr;
bc.n = sm->n;
bc.PL = malloc(15 * sm->n * sizeof(*bc.PL));
if (conf->fmt_flag)
{
assert( sizeof(float)==sizeof(int32_t) );
bc.DP4 = malloc(sm->n * sizeof(int32_t) * 4);
bc.fmt_arr = malloc(sm->n * sizeof(float)); // all fmt_flag fields
if ( conf->fmt_flag&(B2B_INFO_DPR|B2B_FMT_DPR) )
{
// first B2B_MAX_ALLELES fields for total numbers, the rest per-sample
bc.DPR = malloc((sm->n+1)*B2B_MAX_ALLELES*sizeof(int32_t));
for (i=0; i<sm->n; i++)
bcr[i].DPR = bc.DPR + (i+1)*B2B_MAX_ALLELES;
}
}
}
else {
pileup_fp = conf->output_fname? fopen(conf->output_fname, "w") : stdout;
if (pileup_fp == NULL) {
fprintf(stderr, "[%s] failed to write to %s: %s\n", __func__, conf->output_fname, strerror(errno));
exit(1);
}
}
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;
// begin pileup
iter = bam_mplp_init(n, mplp_func, (void**)data);
if ( conf->flag & MPLP_SMART_OVERLAPS ) bam_mplp_init_overlaps(iter);
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);
bcf1_t *bcf_rec = bcf_init1();
int ret;
while ( (ret=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_BCF) {
int total_depth, _ref0, ref16;
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 = seq_nt16_table[_ref0];
bcf_callaux_clean(bca, &bc);
for (i = 0; i < gplp.n; ++i)
bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], ref16, bca, bcr + i);
bc.tid = tid; bc.pos = pos;
bcf_call_combine(gplp.n, bcr, bca, ref16, &bc);
bcf_clear1(bcf_rec);
bcf_call2bcf(&bc, bcf_rec, bcr, conf->fmt_flag, 0, 0);
bcf_write1(bcf_fp, bcf_hdr, bcf_rec);
// call indels; todo: subsampling with total_depth>max_indel_depth instead of ignoring?
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)
{
bcf_callaux_clean(bca, &bc);
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, bca, -1, &bc) >= 0) {
bcf_clear1(bcf_rec);
bcf_call2bcf(&bc, bcf_rec, bcr, conf->fmt_flag, bca, ref);
bcf_write1(bcf_fp, bcf_hdr, bcf_rec);
}
}
} else {
fprintf(pileup_fp, "%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, cnt;
for (j = cnt = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
if (bam_get_qual(p->b)[p->qpos] >= conf->min_baseQ) ++cnt;
}
fprintf(pileup_fp, "\t%d\t", cnt);
if (n_plp[i] == 0) {
fputs("*\t*", pileup_fp);
if (conf->flag & MPLP_PRINT_MAPQ) fputs("\t*", pileup_fp);
if (conf->flag & MPLP_PRINT_POS) fputs("\t*", pileup_fp);
} else {
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
if (bam_get_qual(p->b)[p->qpos] >= conf->min_baseQ)
pileup_seq(pileup_fp, plp[i] + j, pos, ref_len, ref);
}
putc('\t', pileup_fp);
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
int c = bam_get_qual(p->b)[p->qpos];
if (c >= conf->min_baseQ) {
c = c + 33 < 126? c + 33 : 126;
putc(c, pileup_fp);
}
}
if (conf->flag & MPLP_PRINT_MAPQ) {
putc('\t', pileup_fp);
for (j = 0; j < n_plp[i]; ++j) {
const bam_pileup1_t *p = plp[i] + j;
int c = bam_get_qual(p->b)[p->qpos];
if ( c < conf->min_baseQ ) continue;
c = plp[i][j].b->core.qual + 33;
if (c > 126) c = 126;
putc(c, pileup_fp);
}
}
if (conf->flag & MPLP_PRINT_POS) {
putc('\t', pileup_fp);
for (j = 0; j < n_plp[i]; ++j) {
if (j > 0) putc(',', pileup_fp);
fprintf(pileup_fp, "%d", plp[i][j].qpos + 1); // FIXME: printf() is very slow...
}
}
}
}
putc('\n', pileup_fp);
}
}
// clean up
free(bc.tmp.s);
bcf_destroy1(bcf_rec);
if (bcf_fp)
{
hts_close(bcf_fp);
bcf_hdr_destroy(bcf_hdr);
bcf_call_destroy(bca);
free(bc.PL);
free(bc.DP4);
free(bc.DPR);
free(bc.fmt_arr);
free(bcr);
}
if (pileup_fp && conf->output_fname) fclose(pileup_fp);
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);
bam_mplp_destroy(iter);
bam_hdr_destroy(h);
for (i = 0; i < n; ++i) {
sam_close(data[i]->fp);
if (data[i]->iter) hts_itr_destroy(data[i]->iter);
free(data[i]);
}
free(data); free(plp); free(ref); free(n_plp);
return ret;
}
//last is from abc.h
void print_bcf_header(htsFile *fp,bcf_hdr_t *hdr,argStruct *args,kstring_t &buf,const bam_hdr_t *bhdr){
assert(args);
ksprintf(&buf, "##angsdVersion=%s+htslib-%s\n",angsd_version(),hts_version());
bcf_hdr_append(hdr, buf.s);
buf.l = 0;
ksprintf(&buf, "##angsdCommand=");
for (int i=1; i<args->argc; i++)
ksprintf(&buf, " %s", args->argv[i]);
kputc('\n', &buf);
bcf_hdr_append(hdr, buf.s);
buf.l=0;
if(args->ref){
buf.l = 0;
ksprintf(&buf, "##reference=file://%s\n", args->ref);
bcf_hdr_append(hdr,buf.s);
}
if (args->anc){
buf.l = 0;
ksprintf(&buf, "##ancestral=file://%s\n", args->anc);
bcf_hdr_append(hdr,buf.s);
}
// Translate BAM @SQ tags to BCF ##contig tags
// todo: use/write new BAM header manipulation routines, fill also UR, M5
for (int i=0; i<bhdr->n_targets; i++){
buf.l = 0;
ksprintf(&buf, "##contig=<ID=%s,length=%d>", bhdr->target_name[i], bhdr->target_len[i]);
bcf_hdr_append(hdr, buf.s);
}
buf.l = 0;
bcf_hdr_append(hdr,"##ALT=<ID=*,Description=\"Represents allele(s) other than observed.\">");
bcf_hdr_append(hdr,"##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">");
bcf_hdr_append(hdr,"##INFO=<ID=IDV,Number=1,Type=Integer,Description=\"Maximum number of raw reads supporting an indel\">");
bcf_hdr_append(hdr,"##INFO=<ID=IMF,Number=1,Type=Float,Description=\"Maximum fraction of raw reads supporting an indel\">");
bcf_hdr_append(hdr,"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw read depth\">");
bcf_hdr_append(hdr,"##INFO=<ID=VDB,Number=1,Type=Float,Description=\"Variant Distance Bias for filtering splice-site artefacts in RNA-seq data (bigger is better)\",Version=\"3\">");
bcf_hdr_append(hdr,"##INFO=<ID=RPB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias (bigger is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=MQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias (bigger is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=BQB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias (bigger is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=MQSB,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias (bigger is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=NS,Number=1,Type=Integer,Description=\"Number of Samples With Data\">");
bcf_hdr_append(hdr,"##INFO=<ID=RPB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Read Position Bias [CDF] (bigger is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=MQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=BQB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Base Quality Bias [CDF] (bigger is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=MQSB2,Number=1,Type=Float,Description=\"Mann-Whitney U test of Mapping Quality vs Strand Bias [CDF] (bigger is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=SGB,Number=1,Type=Float,Description=\"Segregation based metric.\">");
bcf_hdr_append(hdr,"##INFO=<ID=MQ0F,Number=1,Type=Float,Description=\"Fraction of MQ0 reads (smaller is better)\">");
bcf_hdr_append(hdr,"##INFO=<ID=I16,Number=16,Type=Float,Description=\"Auxiliary tag used for calling, see description of bcf_callret1_t in bam2bcf.h\">");
bcf_hdr_append(hdr,"##INFO=<ID=QS,Number=R,Type=Float,Description=\"Auxiliary tag used for calling\">");
bcf_hdr_append(hdr,"##INFO=<ID=AF,Number=A,Type=Float,Description=\"Minor Allele Frequency\">");
bcf_hdr_append(hdr,"##INFO=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
bcf_hdr_append(hdr,"##INFO=<ID=AD,Number=R,Type=Integer,Description=\"Total allelic depths\">");
bcf_hdr_append(hdr,"##INFO=<ID=ADF,Number=R,Type=Integer,Description=\"Total allelic depths on the forward strand\">");
bcf_hdr_append(hdr,"##INFO=<ID=ADR,Number=R,Type=Integer,Description=\"Total allelic depths on the reverse strand\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=DP4,Number=4,Type=Integer,Description=\"Number of high-quality ref-fwd, ref-reverse, alt-fwd and alt-reverse bases\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=AD,Number=R,Type=Integer,Description=\"Allelic depths\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=ADF,Number=R,Type=Integer,Description=\"Allelic depths on the forward strand\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=ADR,Number=R,Type=Integer,Description=\"Allelic depths on the reverse strand\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=PL,Number=G,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"Number of high-quality bases\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=DV,Number=1,Type=Integer,Description=\"Number of high-quality non-reference bases\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=DPR,Number=R,Type=Integer,Description=\"Number of high-quality bases observed for each allele\">");
bcf_hdr_append(hdr,"##FORMAT=<ID=GL,Number=G,Type=Float,Description=\"scaled Genotype Likelihoods (loglikeratios to the most likely (in log10))\">");
//fprintf(stderr,"samples from sm:%d\n",args->sm->n);
for(int i=0;i<args->sm->n;i++){
bcf_hdr_add_sample(hdr, args->sm->smpl[i]);
}
bcf_hdr_add_sample(hdr, NULL); // to update internal structures
if ( bcf_hdr_write(fp, hdr)!=0 )
fprintf(stderr,"Failed to write bcf\n");
buf.l=0;
}
int main(int argc, char **argv)
{
char *fname = argc>1 ? argv[1] : "/dev/null";
htsFile *fp = hts_open(fname, "w");
bcf_hdr_t *hdr1, *hdr2;
hdr1 = bcf_hdr_init("w");
hdr2 = bcf_hdr_init("w");
// Add two shared and two private annotations
bcf_hdr_append(hdr1, "##contig=<ID=1>");
bcf_hdr_append(hdr1, "##contig=<ID=2>");
bcf_hdr_append(hdr2, "##contig=<ID=2>");
bcf_hdr_append(hdr2, "##contig=<ID=1>");
bcf_hdr_append(hdr1, "##FILTER=<ID=FLT1,Description=\"Filter 1\">");
bcf_hdr_append(hdr1, "##FILTER=<ID=FLT2,Description=\"Filter 2\">");
bcf_hdr_append(hdr1, "##FILTER=<ID=FLT3,Description=\"Filter 3\">");
bcf_hdr_append(hdr2, "##FILTER=<ID=FLT4,Description=\"Filter 4\">");
bcf_hdr_append(hdr2, "##FILTER=<ID=FLT3,Description=\"Filter 3\">");
bcf_hdr_append(hdr2, "##FILTER=<ID=FLT2,Description=\"Filter 2\">");
bcf_hdr_append(hdr1, "##INFO=<ID=INF1,Number=.,Type=Integer,Description=\"Info 1\">");
bcf_hdr_append(hdr1, "##INFO=<ID=INF2,Number=.,Type=Integer,Description=\"Info 2\">");
bcf_hdr_append(hdr1, "##INFO=<ID=INF3,Number=.,Type=Integer,Description=\"Info 3\">");
bcf_hdr_append(hdr2, "##INFO=<ID=INF4,Number=.,Type=Integer,Description=\"Info 4\">");
bcf_hdr_append(hdr2, "##INFO=<ID=INF3,Number=.,Type=Integer,Description=\"Info 3\">");
bcf_hdr_append(hdr2, "##INFO=<ID=INF2,Number=.,Type=Integer,Description=\"Info 2\">");
bcf_hdr_append(hdr1, "##FORMAT=<ID=FMT1,Number=.,Type=Integer,Description=\"FMT 1\">");
bcf_hdr_append(hdr1, "##FORMAT=<ID=FMT2,Number=.,Type=Integer,Description=\"FMT 2\">");
bcf_hdr_append(hdr1, "##FORMAT=<ID=FMT3,Number=.,Type=Integer,Description=\"FMT 3\">");
bcf_hdr_append(hdr2, "##FORMAT=<ID=FMT4,Number=.,Type=Integer,Description=\"FMT 4\">");
bcf_hdr_append(hdr2, "##FORMAT=<ID=FMT3,Number=.,Type=Integer,Description=\"FMT 3\">");
bcf_hdr_append(hdr2, "##FORMAT=<ID=FMT2,Number=.,Type=Integer,Description=\"FMT 2\">");
bcf_hdr_add_sample(hdr1,"SMPL1");
bcf_hdr_add_sample(hdr1,"SMPL2");
bcf_hdr_add_sample(hdr2,"SMPL1");
bcf_hdr_add_sample(hdr2,"SMPL2");
bcf_hdr_sync(hdr1);
bcf_hdr_sync(hdr2);
hdr2 = bcf_hdr_merge(hdr2,hdr1);
bcf_hdr_sync(hdr2);
if ( bcf_hdr_write(fp, hdr2)!=0 ) error("Failed to write to %s\n", fname);
bcf1_t *rec = bcf_init1();
rec->rid = bcf_hdr_name2id(hdr1, "1");
rec->pos = 0;
bcf_update_alleles_str(hdr1, rec, "G,A");
int32_t tmpi[3];
tmpi[0] = bcf_hdr_id2int(hdr1, BCF_DT_ID, "FLT1");
tmpi[1] = bcf_hdr_id2int(hdr1, BCF_DT_ID, "FLT2");
tmpi[2] = bcf_hdr_id2int(hdr1, BCF_DT_ID, "FLT3");
bcf_update_filter(hdr1, rec, tmpi, 3);
tmpi[0] = 1; bcf_update_info_int32(hdr1, rec, "INF1", tmpi, 1);
tmpi[0] = 2; bcf_update_info_int32(hdr1, rec, "INF2", tmpi, 1);
tmpi[0] = 3; bcf_update_info_int32(hdr1, rec, "INF3", tmpi, 1);
tmpi[0] = tmpi[1] = 1; bcf_update_format_int32(hdr1, rec, "FMT1", tmpi, 2);
tmpi[0] = tmpi[1] = 2; bcf_update_format_int32(hdr1, rec, "FMT2", tmpi, 2);
tmpi[0] = tmpi[1] = 3; bcf_update_format_int32(hdr1, rec, "FMT3", tmpi, 2);
bcf_remove_filter(hdr1, rec, bcf_hdr_id2int(hdr1, BCF_DT_ID, "FLT2"), 0);
bcf_update_info_int32(hdr1, rec, "INF2", NULL, 0);
bcf_update_format_int32(hdr1, rec, "FMT2", NULL, 0);
bcf_translate(hdr2, hdr1, rec);
if ( bcf_write(fp, hdr2, rec)!=0 ) error("Faild to write to %s\n", fname);
// Clean
bcf_destroy1(rec);
bcf_hdr_destroy(hdr1);
bcf_hdr_destroy(hdr2);
int ret;
if ( (ret=hts_close(fp)) )
{
fprintf(stderr,"hts_close(%s): non-zero status %d\n",fname,ret);
exit(ret);
}
return 0;
}
int ctx_vcfcov(int argc, char **argv)
{
struct MemArgs memargs = MEM_ARGS_INIT;
const char *out_path = NULL, *out_type = NULL;
uint32_t max_allele_len = 0, max_gt_vars = 0;
char *ref_path = NULL;
bool low_mem = false;
// Arg parsing
char cmd[100];
char shortopts[300];
cmd_long_opts_to_short(longopts, shortopts, sizeof(shortopts));
int c;
size_t i;
// silence error messages from getopt_long
// opterr = 0;
while((c = getopt_long_only(argc, argv, shortopts, longopts, NULL)) != -1) {
cmd_get_longopt_str(longopts, c, cmd, sizeof(cmd));
switch(c) {
case 0: /* flag set */ break;
case 'h': cmd_print_usage(NULL); break;
case 'o': cmd_check(!out_path, cmd); out_path = optarg; break;
case 'O': cmd_check(!out_type, cmd); out_type = optarg; break;
case 'f': cmd_check(!futil_get_force(), cmd); futil_set_force(true); break;
case 'm': cmd_mem_args_set_memory(&memargs, optarg); break;
case 'n': cmd_mem_args_set_nkmers(&memargs, optarg); break;
case 'r': cmd_check(!ref_path, cmd); ref_path = optarg; break;
case 'L': cmd_check(!max_allele_len,cmd); max_allele_len = cmd_uint32(cmd,optarg); break;
case 'N': cmd_check(!max_gt_vars,cmd); max_gt_vars = cmd_uint32(cmd,optarg); break;
case 'M': cmd_check(!low_mem, cmd); low_mem = true; break;
case ':': /* BADARG */
case '?': /* BADCH getopt_long has already printed error */
// cmd_print_usage(NULL);
die("`"CMD" "SUBCMD" -h` for help. Bad option: %s", argv[optind-1]);
default: abort();
}
}
// Defaults for unset values
if(out_path == NULL) out_path = "-";
if(ref_path == NULL) cmd_print_usage("Require a reference (-r,--ref <ref.fa>)");
if(optind+2 > argc) cmd_print_usage("Require VCF and graph files");
if(!max_allele_len) max_allele_len = DEFAULT_MAX_ALLELE_LEN;
if(!max_gt_vars) max_gt_vars = DEFAULT_MAX_GT_VARS;
status("[vcfcov] max allele length: %u; max number of variants: %u",
max_allele_len, max_gt_vars);
// open ref
// index fasta with: samtools faidx ref.fa
faidx_t *fai = fai_load(ref_path);
if(fai == NULL) die("Cannot load ref index: %s / %s.fai", ref_path, ref_path);
// Open input VCF file
const char *vcf_path = argv[optind++];
htsFile *vcffh = hts_open(vcf_path, "r");
if(vcffh == NULL) die("Cannot open VCF file: %s", vcf_path);
bcf_hdr_t *vcfhdr = bcf_hdr_read(vcffh);
if(vcfhdr == NULL) die("Cannot read VCF header: %s", vcf_path);
// Test we can close and reopen files
if(low_mem) {
if((vcffh = hts_open(vcf_path, "r")) == NULL)
die("Cannot re-open VCF file: %s", vcf_path);
if((vcfhdr = bcf_hdr_read(vcffh)) == NULL)
die("Cannot re-read VCF header: %s", vcf_path);
}
//
// Open graph files
//
const size_t num_gfiles = argc - optind;
char **graph_paths = argv + optind;
ctx_assert(num_gfiles > 0);
GraphFileReader *gfiles = ctx_calloc(num_gfiles, sizeof(GraphFileReader));
size_t ncols, ctx_max_kmers = 0, ctx_sum_kmers = 0;
ncols = graph_files_open(graph_paths, gfiles, num_gfiles,
&ctx_max_kmers, &ctx_sum_kmers);
// Check graph + paths are compatible
graphs_gpaths_compatible(gfiles, num_gfiles, NULL, 0, -1);
//
// Decide on memory
//
size_t bits_per_kmer, kmers_in_hash, graph_mem;
bits_per_kmer = sizeof(BinaryKmer)*8 + sizeof(Covg)*8 * ncols;
kmers_in_hash = cmd_get_kmers_in_hash(memargs.mem_to_use,
memargs.mem_to_use_set,
memargs.num_kmers,
memargs.num_kmers_set,
bits_per_kmer,
low_mem ? -1 : (int64_t)ctx_max_kmers,
ctx_sum_kmers,
true, &graph_mem);
cmd_check_mem_limit(memargs.mem_to_use, graph_mem);
//
// Open output file
//
// v=>vcf, z=>compressed vcf, b=>bcf, bu=>uncompressed bcf
int mode = vcf_misc_get_outtype(out_type, out_path);
futil_create_output(out_path);
htsFile *outfh = hts_open(out_path, modes_htslib[mode]);
status("[vcfcov] Output format: %s", hsmodes_htslib[mode]);
// Allocate memory
dBGraph db_graph;
db_graph_alloc(&db_graph, gfiles[0].hdr.kmer_size, ncols, 1, kmers_in_hash,
DBG_ALLOC_COVGS);
//
// Set up tag names
//
// *R => ref, *A => alt
sprintf(kcov_ref_tag, "K%zuR", db_graph.kmer_size); // mean coverage
sprintf(kcov_alt_tag, "K%zuA", db_graph.kmer_size);
// #SAMPLE=<ID=...,K29KCOV=...,K29NK=...,K29RLK>
// - K29_kcov is empirical kmer coverage
// - K29_nkmers is the number of kmers in the sample
// - mean_read_length is the mean read length in bases
char sample_kcov_tag[20], sample_nk_tag[20], sample_rlk_tag[20];
sprintf(sample_kcov_tag, "K%zu_kcov", db_graph.kmer_size); // mean coverage
sprintf(sample_nk_tag, "K%zu_nkmers", db_graph.kmer_size);
sprintf(sample_rlk_tag, "mean_read_length");
//
// Load kmers if we are using --low-mem
//
VcfCovStats st;
memset(&st, 0, sizeof(st));
VcfCovPrefs prefs = {.kcov_ref_tag = kcov_ref_tag,
.kcov_alt_tag = kcov_alt_tag,
.max_allele_len = max_allele_len,
.max_gt_vars = max_gt_vars,
.load_kmers_only = false};
if(low_mem)
{
status("[vcfcov] Loading kmers from VCF+ref");
prefs.load_kmers_only = true;
vcfcov_file(vcffh, vcfhdr, NULL, NULL, vcf_path, fai,
NULL, &prefs, &st, &db_graph);
// Close files
hts_close(vcffh);
bcf_hdr_destroy(vcfhdr);
// Re-open files
if((vcffh = hts_open(vcf_path, "r")) == NULL)
die("Cannot re-open VCF file: %s", vcf_path);
if((vcfhdr = bcf_hdr_read(vcffh)) == NULL)
die("Cannot re-read VCF header: %s", vcf_path);
prefs.load_kmers_only = false;
}
//
// Load graphs
//
GraphLoadingStats gstats;
memset(&gstats, 0, sizeof(gstats));
GraphLoadingPrefs gprefs = graph_loading_prefs(&db_graph);
gprefs.must_exist_in_graph = low_mem;
for(i = 0; i < num_gfiles; i++) {
graph_load(&gfiles[i], gprefs, &gstats);
graph_file_close(&gfiles[i]);
}
ctx_free(gfiles);
hash_table_print_stats(&db_graph.ht);
//
// Set up VCF header / graph matchup
//
size_t *samplehdrids = ctx_malloc(db_graph.num_of_cols * sizeof(size_t));
// Add samples to vcf header
bcf_hdr_t *outhdr = bcf_hdr_dup(vcfhdr);
bcf_hrec_t *hrec;
int sid;
char hdrstr[200];
for(i = 0; i < db_graph.num_of_cols; i++) {
char *sname = db_graph.ginfo[i].sample_name.b;
if((sid = bcf_hdr_id2int(outhdr, BCF_DT_SAMPLE, sname)) < 0) {
bcf_hdr_add_sample(outhdr, sname);
sid = bcf_hdr_id2int(outhdr, BCF_DT_SAMPLE, sname);
}
samplehdrids[i] = sid;
// Add SAMPLE field
hrec = bcf_hdr_get_hrec(outhdr, BCF_HL_STR, "ID", sname, "SAMPLE");
if(hrec == NULL) {
sprintf(hdrstr, "##SAMPLE=<ID=%s,%s=%"PRIu64",%s=%"PRIu64",%s=%zu>", sname,
sample_kcov_tag,
gstats.nkmers[i] ? gstats.sumcov[i] / gstats.nkmers[i] : 0,
sample_nk_tag, gstats.nkmers[i],
sample_rlk_tag, (size_t)db_graph.ginfo[i].mean_read_length);
bcf_hdr_append(outhdr, hdrstr);
}
else {
// mean kcovg
sprintf(hdrstr, "%"PRIu64, gstats.sumcov[i] / gstats.nkmers[i]);
vcf_misc_add_update_hrec(hrec, sample_kcov_tag, hdrstr);
// num kmers
sprintf(hdrstr, "%"PRIu64, gstats.nkmers[i]);
vcf_misc_add_update_hrec(hrec, sample_nk_tag, hdrstr);
// mean read length in kmers
sprintf(hdrstr, "%zu", (size_t)db_graph.ginfo[i].mean_read_length);
vcf_misc_add_update_hrec(hrec, sample_rlk_tag, hdrstr);
}
status("[vcfcov] Colour %zu: %s [VCF column %zu]", i, sname, samplehdrids[i]);
}
// Add genotype format fields
// One field per alternative allele
sprintf(hdrstr, "##FORMAT=<ID=%s,Number=A,Type=Integer,"
"Description=\"Coverage on ref (k=%zu): sum(kmer_covs) / exp_num_kmers\">\n",
kcov_ref_tag, db_graph.kmer_size);
bcf_hdr_append(outhdr, hdrstr);
sprintf(hdrstr, "##FORMAT=<ID=%s,Number=A,Type=Integer,"
"Description=\"Coverage on alt (k=%zu): sum(kmer_covs) / exp_num_kmers\">\n",
kcov_alt_tag, db_graph.kmer_size);
bcf_hdr_append(outhdr, hdrstr);
bcf_hdr_set_version(outhdr, "VCFv4.2");
// Add command string to header
vcf_misc_hdr_add_cmd(outhdr, cmd_get_cmdline(), cmd_get_cwd());
if(bcf_hdr_write(outfh, outhdr) != 0)
die("Cannot write header to: %s", futil_outpath_str(out_path));
status("[vcfcov] Reading %s and adding coverage", vcf_path);
// Reset stats and get coverage
memset(&st, 0, sizeof(st));
vcfcov_file(vcffh, vcfhdr, outfh, outhdr, vcf_path, fai,
samplehdrids, &prefs, &st, &db_graph);
// Print statistics
char ns0[50], ns1[50];
status("[vcfcov] Read %s VCF lines", ulong_to_str(st.nvcf_lines, ns0));
status("[vcfcov] Read %s ALTs", ulong_to_str(st.nalts_read, ns0));
status("[vcfcov] Used %s kmers", ulong_to_str(st.ngt_kmers, ns0));
status("[vcfcov] ALTs used: %s / %s (%.2f%%)",
ulong_to_str(st.nalts_loaded, ns0), ulong_to_str(st.nalts_read, ns1),
st.nalts_read ? (100.0*st.nalts_loaded) / st.nalts_read : 0.0);
status("[vcfcov] ALTs too long (>%ubp): %s / %s (%.2f%%)", max_allele_len,
ulong_to_str(st.nalts_too_long, ns0), ulong_to_str(st.nalts_read, ns1),
st.nalts_read ? (100.0*st.nalts_too_long) / st.nalts_read : 0.0);
status("[vcfcov] ALTs too dense (>%u within %zubp): %s / %s (%.2f%%)",
max_gt_vars, db_graph.kmer_size,
ulong_to_str(st.nalts_no_covg, ns0), ulong_to_str(st.nalts_read, ns1),
st.nalts_read ? (100.0*st.nalts_no_covg) / st.nalts_read : 0.0);
status("[vcfcov] ALTs printed with coverage: %s / %s (%.2f%%)",
ulong_to_str(st.nalts_with_covg, ns0), ulong_to_str(st.nalts_read, ns1),
st.nalts_read ? (100.0*st.nalts_with_covg) / st.nalts_read : 0.0);
status("[vcfcov] Saved to: %s\n", out_path);
ctx_free(samplehdrids);
graph_loading_stats_destroy(&gstats);
bcf_hdr_destroy(vcfhdr);
bcf_hdr_destroy(outhdr);
hts_close(vcffh);
hts_close(outfh);
fai_destroy(fai);
db_graph_dealloc(&db_graph);
return EXIT_SUCCESS;
}
