static void destroy_data(args_t *args)
{
if ( args->flag & CF_CCALL ) ccall_destroy(&args->aux);
else if ( args->flag & CF_MCALL ) mcall_destroy(&args->aux);
else if ( args->flag & CF_QCALL ) qcall_destroy(&args->aux);
int i;
for (i=0; i<args->nsamples; i++) free(args->samples[i]);
if ( args->aux.fams )
{
for (i=0; i<args->aux.nfams; i++) free(args->aux.fams[i].name);
free(args->aux.fams);
}
if ( args->missed_line ) bcf_destroy(args->missed_line);
ploidy_destroy(args->ploidy);
if ( args->gvcf.line ) bcf_destroy(args->gvcf.line);
free(args->gvcf.gt);
free(args->gvcf.dp);
free(args->sex2ploidy);
free(args->sex2ploidy_prev);
free(args->samples);
free(args->samples_map);
free(args->sample2sex);
free(args->aux.ploidy);
bcf_hdr_destroy(args->aux.hdr);
hts_close(args->out_fh);
bcf_sr_destroy(args->aux.srs);
}
static void process(args_t *args)
{
bcf1_t *rec = bcf_sr_get_line(args->sr,0);
bcf_unpack(rec, BCF_UN_ALL);
int i, site_pass = 1;
const uint8_t *smpl_pass = NULL;
if ( args->filter )
{
site_pass = filter_test(args->filter, rec, &smpl_pass);
if ( args->filter_logic & FLT_EXCLUDE ) site_pass = site_pass ? 0 : 1;
}
bcf1_t *out = NULL;
for (i=0; i<rec->n_sample; i++)
{
if ( !args->fh[i] ) continue;
if ( !smpl_pass && !site_pass ) continue;
if ( smpl_pass )
{
int pass = args->filter_logic & FLT_EXCLUDE ? ( smpl_pass[i] ? 0 : 1) : smpl_pass[i];
if ( !pass ) continue;
}
if ( !out ) out = rec_set_info(args, rec);
rec_set_format(args, rec, i, out);
bcf_write(args->fh[i], args->hdr_out, out);
}
if ( out ) bcf_destroy(out);
}
BlockQuantify::BlockQuantifyImpl::~BlockQuantifyImpl()
{
for(auto x : variants)
{
bcf_destroy(x);
}
}
/**
* Returns true if chrom:start1-end1 overlaps with a region in the file.
*/
bool OrderedBCFOverlapMatcher::overlaps_with(std::string& chrom, int32_t start1, int32_t end1)
{
bool overlaps = false;
//moves to new chromosome
if (current_interval.seq!=chrom)
{
buffer.clear();
current_interval.set(chrom);
odr->jump_to_interval(current_interval);
std::cerr << "Jumped to chromosome " << chrom << "\n";
while (odr->read(v))
{
if (bcf_get_end_pos1(v)<start1) continue;
overlaps = overlaps || (bcf_get_pos1(v)<=end1);
buffer.push_back(v);
if (bcf_get_pos1(v)>end1) break;
v = bcf_init();
}
}
else
{
//scythe preceding bed records
std::list<bcf1_t*>::iterator i = buffer.begin();
while (i!=buffer.end())
{
if (bcf_get_end_pos1(*i)<start1)
{
bcf_destroy(*i);
i = buffer.erase(i);
continue;
}
overlaps = (bcf_get_pos1(*i)<=end1);
break;
}
if (!overlaps)
{
if (buffer.empty())
{
while (odr->read(v))
{
if (bcf_get_end_pos1(*i)<start1) continue;
overlaps = overlaps || (bcf_get_pos1(*i)<=end1);
buffer.push_back(v);
if (bcf_get_pos1(v)>end1) break;
v = bcf_init();
}
}
}
}
return overlaps;
};
void call_decomp_destroy(CallDecomp *dc)
{
alignment_free(dc->aln);
needleman_wunsch_free(dc->nw_aligner);
ctx_free(dc->scoring);
bcf_destroy(dc->v);
strbuf_dealloc(&dc->sbuf);
ctx_free(dc);
}
void vcfbuf_destroy(vcfbuf_t *buf)
{
int i;
for (i=0; i<buf->rbuf.m; i++)
if ( buf->vcf[i].rec ) bcf_destroy(buf->vcf[i].rec);
free(buf->vcf);
free(buf->prune.farr);
free(buf->prune.vrec);
free(buf->prune.ac);
free(buf->prune.idx);
free(buf);
}
/**
* Closes files.
*/
void BCFSyncedReader::close()
{
for (size_t i=0; i<nfiles; ++i)
{
bcf_close(files[i]);
bcf_hdr_destroy(hdrs[i]);
bcf_itr_destroy(itrs[i]);
}
while (pool.size()!=0)
{
bcf_destroy(pool.front());
pool.pop_front();
}
}
static int view_vcf(hFILE *hfp, const char *filename)
{
vcfFile *in = hts_hopen(hfp, filename, "r");
if (in == NULL) return 0;
vcfFile *out = dup_stdout("w");
bcf_hdr_t *hdr = bcf_hdr_read(in);
if (show_headers) bcf_hdr_write(out, hdr);
if (mode == view_all) {
bcf1_t *rec = bcf_init();
while (bcf_read(in, hdr, rec) >= 0)
bcf_write(out, hdr, rec);
bcf_destroy(rec);
}
bcf_hdr_destroy(hdr);
hts_close(out);
hts_close(in);
return 1;
}
static void destroy_data(args_t *args)
{
int i;
for (i=0; i<args->nfnames; i++) free(args->fnames[i]);
free(args->fnames);
if ( args->files ) bcf_sr_destroy(args->files);
if ( hts_close(args->out_fh)!=0 ) error("hts_close error\n");
bcf_hdr_destroy(args->out_hdr);
free(args->seen_seq);
free(args->start_pos);
free(args->swap_phase);
for (i=0; i<args->mbuf; i++) bcf_destroy(args->buf[i]);
free(args->buf);
free(args->GTa);
free(args->GTb);
free(args->nmatch);
free(args->nmism);
free(args->phase_qual);
free(args->phase_set);
}
static int query_regions(args_t *args, char *fname, char **regs, int nregs)
{
int i;
htsFile *fp = hts_open(fname,"r");
if ( !fp ) error("Could not read %s\n", fname);
enum htsExactFormat format = hts_get_format(fp)->format;
regidx_t *reg_idx = NULL;
if ( args->targets_fname )
{
reg_idx = regidx_init(args->targets_fname, NULL, NULL, 0, NULL);
if ( !reg_idx ) error("Could not read %s\n", args->targets_fname);
}
if ( format == bcf )
{
htsFile *out = hts_open("-","w");
if ( !out ) error("Could not open stdout\n", fname);
hts_idx_t *idx = bcf_index_load(fname);
if ( !idx ) error("Could not load .csi index of %s\n", fname);
bcf_hdr_t *hdr = bcf_hdr_read(fp);
if ( !hdr ) error("Could not read the header: %s\n", fname);
if ( args->print_header )
bcf_hdr_write(out,hdr);
if ( !args->header_only )
{
bcf1_t *rec = bcf_init();
for (i=0; i<nregs; i++)
{
hts_itr_t *itr = bcf_itr_querys(idx,hdr,regs[i]);
while ( bcf_itr_next(fp, itr, rec) >=0 )
{
if ( reg_idx && !regidx_overlap(reg_idx, bcf_seqname(hdr,rec),rec->pos,rec->pos+rec->rlen-1, NULL) ) continue;
bcf_write(out,hdr,rec);
}
tbx_itr_destroy(itr);
}
bcf_destroy(rec);
}
if ( hts_close(out) ) error("hts_close returned non-zero status for stdout\n");
bcf_hdr_destroy(hdr);
hts_idx_destroy(idx);
}
else if ( format==vcf || format==sam || format==unknown_format )
{
tbx_t *tbx = tbx_index_load(fname);
if ( !tbx ) error("Could not load .tbi/.csi index of %s\n", fname);
kstring_t str = {0,0,0};
if ( args->print_header )
{
while ( hts_getline(fp, KS_SEP_LINE, &str) >= 0 )
{
if ( !str.l || str.s[0]!=tbx->conf.meta_char ) break;
puts(str.s);
}
}
if ( !args->header_only )
{
int nseq;
const char **seq = NULL;
if ( reg_idx ) seq = tbx_seqnames(tbx, &nseq);
for (i=0; i<nregs; i++)
{
hts_itr_t *itr = tbx_itr_querys(tbx, regs[i]);
if ( !itr ) continue;
while (tbx_itr_next(fp, tbx, itr, &str) >= 0)
{
if ( reg_idx && !regidx_overlap(reg_idx,seq[itr->curr_tid],itr->curr_beg,itr->curr_end, NULL) ) continue;
puts(str.s);
}
tbx_itr_destroy(itr);
}
free(seq);
}
free(str.s);
tbx_destroy(tbx);
}
else if ( format==bam )
error("Please use \"samtools view\" for querying BAM files.\n");
if ( reg_idx ) regidx_destroy(reg_idx);
if ( hts_close(fp) ) error("hts_close returned non-zero status: %s\n", fname);
for (i=0; i<nregs; i++) free(regs[i]);
free(regs);
return 0;
}
int main(int argc, char **argv)
{
if (argc < 3) {
fprintf(stderr,"%s <bcf file> <num vars>\n", argv[0]);
return 1;
}
char *fname = argv[1];
uint32_t num_vars = atoi(argv[2]);
htsFile *fp = hts_open(fname,"rb");
bcf_hdr_t *hdr = bcf_hdr_read(fp);
bcf1_t *line = bcf_init1();
int32_t *gt_p = NULL;
uint32_t num_inds = bcf_hdr_nsamples(hdr);
int32_t i, j, k, ntmp = 0, int_i = 0, two_bit_i = 0, sum, t_sum = 0;
uint32_t num_ind_ints = 1 + ((num_inds - 1) / 16);
pri_queue q = priq_new(0);
priority p;
uint32_t *packed_ints = (uint32_t *) calloc(num_ind_ints,
sizeof(uint32_t));
FILE *gt_of = fopen("gt.tmp.packed","wb");
FILE *md_of = fopen("md.tmp.packed","w");
uint32_t *md_index = (uint32_t *) malloc(num_vars * sizeof(uint32_t));
uint32_t md_i = 0;
unsigned long t_bcf_read = 0,
t_bcf_dup = 0,
t_bcf_unpack = 0,
t_bcf_get_genotypes = 0,
t_bcf_hdr_nsamples = 0,
t_q = 0,
t_write = 0,
t_get_md = 0,
t_md_write = 0,
t_pack = 0;
for (i = 0; i < num_vars; ++i) {
sum = 0;
int_i = 0;
two_bit_i = 0;
int r = bcf_read(fp, hdr, line);
// Copy
bcf1_t *t_line = bcf_dup(line);
// Unpack
bcf_unpack(t_line, BCF_UN_ALL);
// Get metadata
size_t len = strlen(bcf_hdr_id2name(hdr, t_line->rid)) +
10 + // max length of pos
strlen(t_line->d.id) +
strlen(t_line->d.allele[0]) +
strlen(t_line->d.allele[1]) +
4; //tabs
char *md = (char *) malloc(len * sizeof(char));
sprintf(md, "%s\t%d\t%s\t%s\t%s",
bcf_hdr_id2name(hdr, t_line->rid),
t_line->pos + 1,
t_line->d.id,
t_line->d.allele[0],
t_line->d.allele[1]);
// Write metadata
md_i += strlen(md);
md_index[i] = md_i;
fprintf(md_of, "%s", md);
// Get gentotypes
uint32_t num_gts_per_sample = bcf_get_genotypes(hdr,
t_line,
>_p,
&ntmp);
num_gts_per_sample /= num_inds;
int32_t *gt_i = gt_p;
// Pack genotypes
for (j = 0; j < num_inds; ++j) {
uint32_t gt = 0;
for (k = 0; k < num_gts_per_sample; ++k) {
gt += bcf_gt_allele(gt_i[k]);
}
packed_ints[int_i] += gt << (30 - 2*two_bit_i);
two_bit_i += 1;
if (two_bit_i == 16) {
two_bit_i = 0;
int_i += 1;
}
sum += gt;
gt_i += num_gts_per_sample;
}
// Get a priority for the variant based on the sum and number of
// leading zeros
p.sum = sum;
uint32_t prefix_len = 0;
j = 0;
while ((j < num_ind_ints) && (packed_ints[j] == 0)){
prefix_len += 32;
j += 1;
}
if (j < num_ind_ints)
prefix_len += nlz1(packed_ints[j]);
// Push it into the q
p.len = prefix_len;
int *j = (int *) malloc (sizeof(int));
j[0] = i;
priq_push(q, j, p);
// Write to file
fwrite(packed_ints, sizeof(uint32_t), num_ind_ints,gt_of);
memset(packed_ints, 0, num_ind_ints*sizeof(uint32_t));
t_sum += sum;
bcf_destroy(t_line);
free(md);
}
fclose(gt_of);
fclose(md_of);
md_of = fopen("md.tmp.packed","r");
FILE *md_out = fopen("md.bim","w");
gt_of = fopen("gt.tmp.packed","rb");
FILE *s_gt_of = fopen("s.gt.tmp.packed","wb");
// Get variants in order and rewrite a variant-major sorted matrix
while ( priq_top(q, &p) != NULL ) {
int *d = priq_pop(q, &p);
uint32_t start = 0;
if (*d != 0)
start = md_index[*d - 1];
uint32_t len = md_index[*d] - start;
fseek(md_of, start*sizeof(char), SEEK_SET);
char buf[len+1];
fread(buf, sizeof(char), len, md_of);
buf[len] = '\0';
fseek(gt_of, (*d)*num_ind_ints*sizeof(uint32_t), SEEK_SET);
fread(packed_ints, sizeof(uint32_t), num_ind_ints, gt_of);
fwrite(packed_ints, sizeof(uint32_t), num_ind_ints,s_gt_of);
fprintf(md_out, "%s\n", buf);
}
fclose(md_out);
fclose(md_of);
fclose(gt_of);
fclose(s_gt_of);
/*
* In a packed-int variant-major matrix there will be a num_vars
* number of rows, and a num_inds number of values packed into
* num_inds_ints number of intergers. For examples, 16 rows of 16 values
* will be 16 ints, where each int encodes 16 values.
*
*/
uint32_t num_var_ints = 1 + ((num_vars - 1) / 16);
uint32_t *I_data = (uint32_t *)calloc(num_var_ints*16,sizeof(uint32_t));
uint32_t **I = (uint32_t **)malloc(16*sizeof(uint32_t*));
for (i = 0; i < 16; ++i)
I[i] = I_data + i*num_var_ints;
uint32_t I_i = 0, I_int_i = 0;
uint32_t v;
s_gt_of = fopen("s.gt.tmp.packed","rb");
FILE *rs_gt_of = fopen("r.s.gt.tmp.packed","wb");
// Write these to values to that this is a well-formed uncompressed
// packed int binary file (ubin) file
fwrite(&num_vars, sizeof(uint32_t), 1, rs_gt_of);
fwrite(&num_inds, sizeof(uint32_t), 1, rs_gt_of);
/*
* we need to loop over the columns in the v-major file.
* There are num_vars rows, and num_ind_ints 16-ind packed columns
*
* In this loop :
* i: cols in var-major form, rows in ind-major form
* j: rows in var-major form, cols in ind-major form
*/
uint32_t num_inds_to_write = num_inds;
for (i = 0; i < num_ind_ints; ++i) { // loop over each int col
for (j = 0; j < num_vars; ++j) { // loop over head row in that col
// skip to the value at the row/col
fseek(s_gt_of,
j*num_ind_ints*sizeof(uint32_t) + //row
i*sizeof(uint32_t), //col
SEEK_SET);
fread(&v, sizeof(uint32_t), 1, s_gt_of);
// one int corresponds to a col of 16 two-bit values
// two_bit_i will move across the cols
for (two_bit_i = 0; two_bit_i < 16; ++two_bit_i) {
I[two_bit_i][I_i] += ((v >> (30 - 2*two_bit_i)) & 3) <<
(30 - 2*I_int_i);
}
I_int_i += 1;
if (I_int_i == 16) {
I_i += 1;
I_int_i = 0;
}
}
// When we are at the end of the file, and the number of lines
// is not a factor of 16, only write out the lines that contain values
if (num_inds_to_write >= 16) {
fwrite(I_data,
sizeof(uint32_t),
num_var_ints*16,
rs_gt_of);
num_inds_to_write -= 16;
} else {
fwrite(I_data,
sizeof(uint32_t),
num_var_ints*num_inds_to_write,
rs_gt_of);
}
memset(I_data, 0, num_var_ints*16*sizeof(uint32_t));
I_int_i = 0;
I_i = 0;
}
fclose(s_gt_of);
fclose(rs_gt_of);
free(md_index);
free(packed_ints);
}
static void init_data(args_t *args)
{
bcf1_t *line = NULL;
// With phased concat, the chunks overlap and come in the right order. To
// avoid opening all files at once, store start positions to recognise need
// for the next one. This way we can keep only two open chunks at once.
if ( args->phased_concat )
{
args->start_pos = (int*) malloc(sizeof(int)*args->nfnames);
line = bcf_init();
}
kstring_t str = {0,0,0};
int i, prev_chrid = -1;
for (i=0; i<args->nfnames; i++)
{
htsFile *fp = hts_open(args->fnames[i], "r"); if ( !fp ) error("Failed to open: %s\n", args->fnames[i]);
bcf_hdr_t *hdr = bcf_hdr_read(fp); if ( !hdr ) error("Failed to parse header: %s\n", args->fnames[i]);
args->out_hdr = bcf_hdr_merge(args->out_hdr,hdr);
if ( bcf_hdr_nsamples(hdr) != bcf_hdr_nsamples(args->out_hdr) )
error("Different number of samples in %s. Perhaps \"bcftools merge\" is what you are looking for?\n", args->fnames[i]);
int j;
for (j=0; j<bcf_hdr_nsamples(hdr); j++)
if ( strcmp(args->out_hdr->samples[j],hdr->samples[j]) )
error("Different sample names in %s. Perhaps \"bcftools merge\" is what you are looking for?\n", args->fnames[i]);
if ( args->phased_concat )
{
int ret = bcf_read(fp, hdr, line);
if ( ret!=0 ) args->start_pos[i] = -2; // empty file
else
{
int chrid = bcf_hdr_id2int(args->out_hdr,BCF_DT_CTG,bcf_seqname(hdr,line));
args->start_pos[i] = chrid==prev_chrid ? line->pos : -1;
prev_chrid = chrid;
}
}
bcf_hdr_destroy(hdr);
hts_close(fp);
}
free(str.s);
if ( line ) bcf_destroy(line);
args->seen_seq = (int*) calloc(args->out_hdr->n[BCF_DT_CTG],sizeof(int));
if ( args->phased_concat )
{
bcf_hdr_append(args->out_hdr,"##FORMAT=<ID=PQ,Number=1,Type=Integer,Description=\"Phasing Quality (bigger is better)\">");
bcf_hdr_append(args->out_hdr,"##FORMAT=<ID=PS,Number=1,Type=Integer,Description=\"Phase Set\">");
}
if (args->record_cmd_line) bcf_hdr_append_version(args->out_hdr, args->argc, args->argv, "bcftools_concat");
args->out_fh = hts_open(args->output_fname,hts_bcf_wmode(args->output_type));
if ( args->out_fh == NULL ) error("Can't write to \"%s\": %s\n", args->output_fname, strerror(errno));
if ( args->n_threads ) hts_set_threads(args->out_fh, args->n_threads);
bcf_hdr_write(args->out_fh, args->out_hdr);
if ( args->allow_overlaps )
{
args->files = bcf_sr_init();
args->files->require_index = 1;
if ( args->regions_list )
{
if ( bcf_sr_set_regions(args->files, args->regions_list, args->regions_is_file)<0 )
error("Failed to read the regions: %s\n", args->regions_list);
}
if ( args->remove_dups )
{
if ( !strcmp(args->remove_dups,"snps") ) args->files->collapse |= COLLAPSE_SNPS;
else if ( !strcmp(args->remove_dups,"indels") ) args->files->collapse |= COLLAPSE_INDELS;
else if ( !strcmp(args->remove_dups,"both") ) args->files->collapse |= COLLAPSE_SNPS | COLLAPSE_INDELS;
else if ( !strcmp(args->remove_dups,"any") ) args->files->collapse |= COLLAPSE_ANY;
else if ( !strcmp(args->remove_dups,"all") ) args->files->collapse |= COLLAPSE_ANY;
else if ( !strcmp(args->remove_dups,"none") ) args->files->collapse = COLLAPSE_NONE;
else error("The -D string \"%s\" not recognised.\n", args->remove_dups);
}
for (i=0; i<args->nfnames; i++)
if ( !bcf_sr_add_reader(args->files,args->fnames[i]) ) error("Failed to open %s: %s\n", args->fnames[i],bcf_sr_strerror(args->files->errnum));
}
else if ( args->phased_concat )
{
// Remove empty files from the list
int nok = 0;
while (1)
{
while ( nok<args->nfnames && args->start_pos[nok]!=-2 ) nok++;
if ( nok==args->nfnames ) break;
i = nok;
while ( i<args->nfnames && args->start_pos[i]==-2 ) i++;
if ( i==args->nfnames ) break;
int tmp = args->start_pos[nok]; args->start_pos[nok] = args->start_pos[i]; args->start_pos[i] = tmp;
char *str = args->fnames[nok]; args->fnames[nok] = args->fnames[i]; args->fnames[i] = str;
}
for (i=nok; i<args->nfnames; i++) free(args->fnames[i]);
args->nfnames = nok;
for (i=1; i<args->nfnames; i++)
if ( args->start_pos[i-1]!=-1 && args->start_pos[i]!=-1 && args->start_pos[i]<args->start_pos[i-1] )
error("The files not in ascending order: %d in %s, %d in %s\n", args->start_pos[i-1]+1,args->fnames[i-1],args->start_pos[i]+1,args->fnames[i]);
args->prev_chr = -1;
args->swap_phase = (int*) calloc(bcf_hdr_nsamples(args->out_hdr),sizeof(int));
args->nmatch = (int*) calloc(bcf_hdr_nsamples(args->out_hdr),sizeof(int));
args->nmism = (int*) calloc(bcf_hdr_nsamples(args->out_hdr),sizeof(int));
args->phase_qual = (int32_t*) malloc(bcf_hdr_nsamples(args->out_hdr)*sizeof(int32_t));
args->phase_set = (int32_t*) malloc(bcf_hdr_nsamples(args->out_hdr)*sizeof(int32_t));
args->files = bcf_sr_init();
args->files->require_index = 1;
args->ifname = 0;
}
}
/**
* Flush records.
*/
void BCFSingleGenotypingBufferedReader::flush(bam_hdr_t *h, bam1_t *s, bool flush_all)
{
if (flush_all)
{
//read all the remaining from the reference genotyping file
bcf1_t *v = bcf_init();
while (odr->read(v))
{
int32_t vtype = vm->classify_variant(odr->hdr, v, variant);
GenotypingRecord* g = create_genotyping_record(odr->hdr, v, 2, variant);
buffer.push_back(g);
v = bcf_init();
}
bcf_destroy(v);
GenotypingRecord* g;
while (!buffer.empty())
{
g = buffer.front();
// genotype_and_print(g);
delete g;
buffer.pop_front();
}
}
else
{
//std::cerr << "partial flush\n";
uint32_t tid = bam_get_tid(s);
GenotypingRecord* g;
while (!buffer.empty())
{
g = buffer.front();
if (tid==g->rid)
{
if (bam_get_pos1(s) > g->end1)
{
// genotype_and_print(g);
delete g;
buffer.pop_front();
}
else
{
return;
}
}
else if (tid>g->rid)
{
// genotype_and_print(g);
delete g;
buffer.pop_front();
}
else
{
return;
}
}
}
}
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;
}
int main(int argc, char **argv)
{
if ( argc == 1 )
error("Usage : bed_annos -c config.json -O z -o output.vcf.gz input.vcf.gz");
int i;
for ( i = 1; i < argc; ) {
const char *a = argv[i++];
const char **var = 0;
if ( strcmp(a, "-c") == 0 )
var = &json_fname;
else if ( strcmp(a, "-O") == 0 )
var = &output_fname_type;
else if ( strcmp(a, "-o") == 0 )
var = &output_fname;
if ( var != 0 ) {
if ( i == argc )
error("Missing an argument after %s", a);
*var = argv[i++];
continue;
}
if ( input_fname == 0 ) {
input_fname = a;
continue;
}
error("Unknown argument : %s.", a);
}
struct vcfanno_config *con = vcfanno_config_init();
if ( vcfanno_load_config(con, json_fname) != 0 )
error("Failed to load configure file. %s : %s", json_fname, strerror(errno));
vcfanno_config_debug(con);
if ( con->beds.n_beds == 0)
error("No bed database specified.");
if ( input_fname == 0 && (!isatty(fileno(stdin))) )
input_fname = "-";
if ( input_fname == 0 )
error("No input file.");
int out_type = FT_VCF;
if ( output_fname_type != 0 ) {
switch (output_fname_type[0]) {
case 'b':
out_type = FT_BCF_GZ; break;
case 'u':
out_type = FT_BCF; break;
case 'z':
out_type = FT_VCF_GZ; break;
case 'v':
out_type = FT_VCF; break;
default :
error("The output type \"%d\" not recognised\n", out_type);
};
}
htsFile *fp = hts_open(input_fname, "r");
if ( fp == NULL )
error("Failed to open %s : %s.", input_fname, strerror(errno));
htsFormat type = *hts_get_format(fp);
if ( type.format != vcf && type.format != bcf )
error("Unsupported input format. %s", input_fname);
bcf_hdr_t *hdr = bcf_hdr_read(fp);
if ( hdr == NULL )
error("Failed to parse header.");
bcf_hdr_t *hdr_out = bcf_hdr_dup(hdr);
htsFile *fout = output_fname == 0 ? hts_open("-", hts_bcf_wmode(out_type)) : hts_open(output_fname, hts_bcf_wmode(out_type));
struct beds_options opts = { .beds_is_inited = 0,};
beds_options_init(&opts);
opts.hdr_out = hdr_out;
for ( i = 0; i < con->beds.n_beds; ++i ) {
beds_database_add(&opts, con->beds.files[i].fname, con->beds.files[i].columns);
}
bcf_hdr_write(fout, hdr_out);
bcf1_t *line = bcf_init();
while ( bcf_read(fp, hdr, line) == 0 ) {
anno_beds_core(&opts, line);
bcf_write(fout, hdr_out, line);
}
bcf_destroy(line);
bcf_hdr_destroy(hdr);
bcf_hdr_destroy(hdr_out);
beds_options_destroy(&opts);
hts_close(fp);
hts_close(fout);
return 0;
}
/**
* Destructor.
*/
GenotypingRecord::~GenotypingRecord()
{
if (v) bcf_destroy(v);
}
static void concat(args_t *args)
{
int i;
if ( args->phased_concat ) // phased concat
{
// keep only two open files at a time
while ( args->ifname < args->nfnames )
{
int new_file = 0;
while ( args->files->nreaders < 2 && args->ifname < args->nfnames )
{
if ( !bcf_sr_add_reader(args->files,args->fnames[args->ifname]) ) error("Failed to open %s: %s\n", args->fnames[args->ifname],bcf_sr_strerror(args->files->errnum));
new_file = 1;
args->ifname++;
if ( args->start_pos[args->ifname-1]==-1 ) break; // new chromosome, start with only one file open
if ( args->ifname < args->nfnames && args->start_pos[args->ifname]==-1 ) break; // next file starts on a different chromosome
}
// is there a line from the previous run? Seek the newly opened reader to that position
int seek_pos = -1;
int seek_chr = -1;
if ( bcf_sr_has_line(args->files,0) )
{
bcf1_t *line = bcf_sr_get_line(args->files,0);
bcf_sr_seek(args->files, bcf_seqname(args->files->readers[0].header,line), line->pos);
seek_pos = line->pos;
seek_chr = bcf_hdr_name2id(args->out_hdr, bcf_seqname(args->files->readers[0].header,line));
}
else if ( new_file )
bcf_sr_seek(args->files,NULL,0); // set to start
int nret;
while ( (nret = bcf_sr_next_line(args->files)) )
{
if ( !bcf_sr_has_line(args->files,0) ) // no input from the first reader
{
// We are assuming that there is a perfect overlap, sites which are not present in both files are dropped
if ( ! bcf_sr_region_done(args->files,0) ) continue;
phased_flush(args);
bcf_sr_remove_reader(args->files, 0);
}
// Get a line to learn about current position
for (i=0; i<args->files->nreaders; i++)
if ( bcf_sr_has_line(args->files,i) ) break;
bcf1_t *line = bcf_sr_get_line(args->files,i);
// This can happen after bcf_sr_seek: indel may start before the coordinate which we seek to.
if ( seek_chr>=0 && seek_pos>line->pos && seek_chr==bcf_hdr_name2id(args->out_hdr, bcf_seqname(args->files->readers[i].header,line)) ) continue;
seek_pos = seek_chr = -1;
// Check if the position overlaps with the next, yet unopened, reader
int must_seek = 0;
while ( args->ifname < args->nfnames && args->start_pos[args->ifname]!=-1 && line->pos >= args->start_pos[args->ifname] )
{
must_seek = 1;
if ( !bcf_sr_add_reader(args->files,args->fnames[args->ifname]) ) error("Failed to open %s: %s\n", args->fnames[args->ifname],bcf_sr_strerror(args->files->errnum));
args->ifname++;
}
if ( must_seek )
{
bcf_sr_seek(args->files, bcf_seqname(args->files->readers[i].header,line), line->pos);
seek_pos = line->pos;
seek_chr = bcf_hdr_name2id(args->out_hdr, bcf_seqname(args->files->readers[i].header,line));
continue;
}
// We are assuming that there is a perfect overlap, sites which are not present in both files are dropped
if ( args->files->nreaders>1 && !bcf_sr_has_line(args->files,1) && !bcf_sr_region_done(args->files,1) ) continue;
phased_push(args, bcf_sr_get_line(args->files,0), args->files->nreaders>1 ? bcf_sr_get_line(args->files,1) : NULL);
}
if ( args->files->nreaders )
{
phased_flush(args);
while ( args->files->nreaders )
bcf_sr_remove_reader(args->files, 0);
}
}
}
else if ( args->files ) // combining overlapping files, using synced reader
{
while ( bcf_sr_next_line(args->files) )
{
for (i=0; i<args->files->nreaders; i++)
{
bcf1_t *line = bcf_sr_get_line(args->files,i);
if ( !line ) continue;
bcf_translate(args->out_hdr, args->files->readers[i].header, line);
bcf_write1(args->out_fh, args->out_hdr, line);
if ( args->remove_dups ) break;
}
}
}
else // concatenating
{
kstring_t tmp = {0,0,0};
int prev_chr_id = -1, prev_pos;
bcf1_t *line = bcf_init();
for (i=0; i<args->nfnames; i++)
{
htsFile *fp = hts_open(args->fnames[i], "r"); if ( !fp ) error("Failed to open: %s\n", args->fnames[i]);
bcf_hdr_t *hdr = bcf_hdr_read(fp); if ( !hdr ) error("Failed to parse header: %s\n", args->fnames[i]);
if ( !fp->is_bin && args->output_type&FT_VCF )
{
line->max_unpack = BCF_UN_STR;
// if VCF is on both input and output, avoid VCF to BCF conversion
while ( hts_getline(fp, KS_SEP_LINE, &fp->line) >=0 )
{
char *str = fp->line.s;
while ( *str && *str!='\t' ) str++;
tmp.l = 0;
kputsn(fp->line.s,str-fp->line.s,&tmp);
int chr_id = bcf_hdr_name2id(args->out_hdr, tmp.s);
if ( chr_id<0 ) error("The sequence \"%s\" not defined in the header: %s\n(Quick workaround: index the file.)\n", tmp.s, args->fnames[i]);
if ( prev_chr_id!=chr_id )
{
prev_pos = -1;
if ( args->seen_seq[chr_id] )
error("\nThe chromosome block %s is not contiguous, consider running with -a.\n", tmp.s);
}
char *end;
int pos = strtol(str+1,&end,10) - 1;
if ( end==str+1 ) error("Could not parse line: %s\n", fp->line.s);
if ( prev_pos > pos )
error("The chromosome block %s is not sorted, consider running with -a.\n", tmp.s);
args->seen_seq[chr_id] = 1;
prev_chr_id = chr_id;
if ( vcf_write_line(args->out_fh, &fp->line)!=0 ) error("Failed to write %d bytes\n", fp->line.l);
}
}
else
{
// BCF conversion is required
line->max_unpack = 0;
while ( bcf_read(fp, hdr, line)==0 )
{
bcf_translate(args->out_hdr, hdr, line);
if ( prev_chr_id!=line->rid )
{
prev_pos = -1;
if ( args->seen_seq[line->rid] )
error("\nThe chromosome block %s is not contiguous, consider running with -a.\n", bcf_seqname(args->out_hdr, line));
}
if ( prev_pos > line->pos )
error("The chromosome block %s is not sorted, consider running with -a.\n", bcf_seqname(args->out_hdr, line));
args->seen_seq[line->rid] = 1;
prev_chr_id = line->rid;
if ( bcf_write(args->out_fh, args->out_hdr, line)!=0 ) error("Failed to write\n");
}
}
bcf_hdr_destroy(hdr);
hts_close(fp);
}
bcf_destroy(line);
free(tmp.s);
}
}
/**
* Collects sufficient statistics from read for variants to be genotyped.
*
* The VCF records in the buffer must never occur before
*/
void BCFSingleGenotypingBufferedReader::process_read(bam_hdr_t *h, bam1_t *s)
{
//wrap bam1_t in AugmentBAMRecord
as.initialize(h, s);
uint32_t tid = bam_get_tid(s);
uint32_t beg1 = as.beg1;
uint32_t end1 = as.end1;
//collect statistics for variant records that are in the buffer and overlap with the read
GenotypingRecord* g;
for (std::list<GenotypingRecord*>::iterator i=buffer.begin(); i!=buffer.end(); ++i)
{
g = *i;
// std::cerr << g->pos1 << " " << g->beg1 << " " << g->end1 << " ";
//same chromosome
if (tid==g->rid)
{
if (end1 < g->beg1)
{
//can terminate
return;
}
else if (beg1 > g->end1)
{
//this should not occur if the buffer was flushed before invoking process read
continue;
}
//else if (beg1 <= g->beg1 && g->end1 <= end1)
else if (beg1 <= g->pos1 && g->pos1 <= end1)
{
// collect_sufficient_statistics(*i, as);
}
else
{
//other types of overlap, just ignore
}
// std::cerr << "\n";
}
//prior chromosome
else if (tid<g->rid)
{
//this should not occur if the buffer was flushed before invoking process read
return;
}
//latter chromosome
else if (tid>g->rid)
{
//in case if the buffer has a VCF record later in the list which matches it
continue;
}
}
//you will only reach here if a read occurs after or overlaps the last record in the buffer
//adding new VCF records and collecting statistics if necessary
bcf1_t *v = bcf_init();
while (odr->read(v))
{
int32_t vtype = vm->classify_variant(odr->hdr, v, variant);
g = create_genotyping_record(odr->hdr, v, 2, variant);
buffer.push_back(g);
if (tid==g->rid)
{
//if (end1>=g->beg1 && pos1<=g->end1)
if (beg1 <= g->pos1 && g->pos1 <= end1)
{
// collect_sufficient_statistics(g, as);
}
}
//VCF record occurs after the read
if (tid < g->rid || end1 < g->beg1)
{
return;
}
else
{
v = bcf_init();
}
}
//this means end of file
bcf_destroy(v);
}
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;
}
static void reheader_bcf(args_t *args, int is_compressed)
{
htsFile *fp = hts_open(args->fname, "r"); if ( !fp ) error("Failed to open: %s\n", args->fname);
bcf_hdr_t *hdr = bcf_hdr_read(fp); if ( !hdr ) error("Failed to read the header: %s\n", args->fname);
kstring_t htxt = {0,0,0};
int hlen;
htxt.s = bcf_hdr_fmt_text(hdr, 1, &hlen);
htxt.l = hlen;
int i, nsamples = 0;
char **samples = NULL;
if ( args->samples_fname )
samples = hts_readlines(args->samples_fname, &nsamples);
if ( args->header_fname )
{
free(htxt.s); htxt.s = NULL; htxt.l = htxt.m = 0;
read_header_file(args->header_fname, &htxt);
}
if ( samples )
{
set_samples(samples, nsamples, &htxt);
for (i=0; i<nsamples; i++) free(samples[i]);
free(samples);
}
bcf_hdr_t *hdr_out = bcf_hdr_init("r");
bcf_hdr_parse(hdr_out, htxt.s);
if ( args->header_fname ) hdr_out = strip_header(hdr, hdr_out);
// write the header and the body
htsFile *fp_out = hts_open("-",is_compressed ? "wb" : "wbu");
bcf_hdr_write(fp_out, hdr_out);
bcf1_t *rec = bcf_init();
while ( bcf_read(fp, hdr, rec)==0 )
{
// sanity checking, this slows things down. Make it optional?
bcf_unpack(rec, BCF_UN_ALL);
if ( rec->rid >= hdr_out->n[BCF_DT_CTG] || strcmp(bcf_hdr_int2id(hdr,BCF_DT_CTG,rec->rid),bcf_hdr_int2id(hdr_out,BCF_DT_CTG,rec->rid)) )
error("The CHROM is not defined: \"%s\"\n", bcf_hdr_int2id(hdr,BCF_DT_CTG,rec->rid));
for (i=0; i<rec->d.n_flt; i++)
{
int id = rec->d.flt[i];
if ( id >= hdr_out->n[BCF_DT_ID] ) break;
if ( !bcf_hdr_idinfo_exists(hdr_out,BCF_HL_FLT,id) ) break;
if ( strcmp(hdr->id[BCF_DT_ID][id].key,hdr_out->id[BCF_DT_ID][id].key) )
error("FIXME: Broken FILTER ids: %s vs %s\n", hdr->id[BCF_DT_ID][id].key,hdr_out->id[BCF_DT_ID][id].key);
}
if ( i!=rec->d.n_flt )
error("The FILTER is not defined: \"%s\"\n", bcf_hdr_int2id(hdr,BCF_DT_ID,rec->d.flt[i]));
for (i=0; i<rec->n_info; i++)
{
int id = rec->d.info[i].key;
if ( id >= hdr_out->n[BCF_DT_ID] ) break;
if ( !hdr_out->id[BCF_DT_ID][id].key ) break;
if ( !bcf_hdr_idinfo_exists(hdr_out,BCF_HL_INFO,id) ) break;
if ( strcmp(hdr->id[BCF_DT_ID][id].key,hdr_out->id[BCF_DT_ID][id].key) )
error("FIXME: Broken INFO ids: %s vs %s\n", hdr->id[BCF_DT_ID][id].key,hdr_out->id[BCF_DT_ID][id].key);
}
if ( i!=rec->n_info )
error("The INFO tag is not defined: \"%s\"\n", bcf_hdr_int2id(hdr,BCF_DT_ID,rec->d.info[i].key));
for (i=0; i<rec->n_fmt; i++)
{
int id = rec->d.fmt[i].id;
if ( id >= hdr_out->n[BCF_DT_ID] ) break;
if ( !hdr_out->id[BCF_DT_ID][id].key ) break;
if ( !bcf_hdr_idinfo_exists(hdr_out,BCF_HL_FMT,id) ) break;
if ( strcmp(hdr->id[BCF_DT_ID][id].key,hdr_out->id[BCF_DT_ID][id].key) )
error("FIXME: Broken FORMAT ids: %s vs %s\n", hdr->id[BCF_DT_ID][id].key,hdr_out->id[BCF_DT_ID][id].key);
}
if ( i!=rec->n_fmt )
error("The FORMAT tag is not defined: \"%s\"\n", bcf_hdr_int2id(hdr,BCF_DT_ID,rec->d.fmt[i].id));
bcf_write(fp_out,hdr_out,rec);
}
bcf_destroy(rec);
free(htxt.s);
hts_close(fp_out);
hts_close(fp);
bcf_hdr_destroy(hdr_out);
bcf_hdr_destroy(hdr);
}
