// add ROC decision point
void BlockQuantify::addROCValue(std::string const & roc_identifier,
roc::DecisionType dt,
double level,
uint64_t n,
bcf1_t * v)
{
// add observation to a roc
auto observe = [this, level, dt, n, v](std::string const & name, bool f) {
roc::DecisionType final_dt = dt;
if(f)
{
if(dt == roc::DecisionType::TP)
{
// filter-failed TPs become FNs
final_dt = roc::DecisionType::FN;
}
else if(dt == roc::DecisionType::TP2)
{
// filter-failed TPs become FNs
final_dt = roc::DecisionType::FN2;
}
else if(dt != roc::DecisionType::FN)
{
// filter-failed FPs / UNKs become Ns
final_dt = roc::DecisionType::N;
}
}
uint64_t flags = 0;
switch(final_dt)
{
case roc::DecisionType::FN:
case roc::DecisionType::TP:
{
const std::string bk_truth = bcfhelpers::getFormatString(_impl->hdr, v, "BK", 0, ".");
const std::string bi_truth = bcfhelpers::getFormatString(_impl->hdr, v, "BI", 0, ".");
const std::string blt_truth = bcfhelpers::getFormatString(_impl->hdr, v, "BLT", 0, ".");
flags = roc::makeObservationFlags(bk_truth, bi_truth, blt_truth);
break;
}
case roc::DecisionType::FP:
case roc::DecisionType::UNK:
case roc::DecisionType::TP2:
case roc::DecisionType::FN2:
{
const std::string bk_query = bcfhelpers::getFormatString(_impl->hdr, v, "BK", 1, ".");
const std::string bi_query = bcfhelpers::getFormatString(_impl->hdr, v, "BI", 1, ".");
const std::string blt_query = bcfhelpers::getFormatString(_impl->hdr, v, "BLT", 1, ".");
flags = roc::makeObservationFlags(bk_query, bi_query, blt_query);
break;
}
default:
break;
}
auto it = _impl->rocs.find(name);
if(it == _impl->rocs.end())
{
it = _impl->rocs.insert(std::make_pair(name, roc::Roc())).first;
}
// make sure FN and N always come first
if( ( final_dt == roc::DecisionType::FN
|| final_dt == roc::DecisionType::FN2
|| final_dt == roc::DecisionType::N )
&& level == 0
)
{
it->second.add(roc::Observation{std::numeric_limits<double>::min(), final_dt, n, flags});
}
else
{
it->second.add(roc::Observation{level, final_dt, n, flags});
}
};
bcf_unpack(v, BCF_UN_FLT);
bool fail = false; // fails any of the non-blocked filters
bool fail_any = false; // fails any filter
for(int j = 0; j < v->d.n_flt; ++j)
{
std::string filter = "PASS";
int k = v->d.flt[j];
if(k >= 0)
{
filter = bcf_hdr_int2id(_impl->hdr, BCF_DT_ID, v->d.flt[j]);
}
if(filter != "PASS" && filter != "")
{
if(!_impl->filters_to_ignore.count("*") && !_impl->filters_to_ignore.count(filter))
{
fail = true;
observe("f:" + roc_identifier + ":" + filter, false);
}
else
{
observe("f:" + roc_identifier + ":SEL_IGN_" + filter, false);
}
fail_any = true;
}
}
observe("a:" + roc_identifier + ":PASS", fail_any);
// selectively-filtered ROCs
if(!_impl->filters_to_ignore.empty())
{
observe("a:" + roc_identifier + ":SEL", fail);
}
observe("a:" + roc_identifier + ":ALL", false);
const std::string regions = bcfhelpers::getInfoString(_impl->hdr, v, "Regions", "");
if(!regions.empty() && regions != "CONF")
{
std::vector<std::string> rs;
stringutil::split(regions, rs, ",");
for(auto const & r : rs)
{
if(r == "CONF")
{
continue;
}
observe("s|" + r + ":" + roc_identifier + ":PASS", fail_any);
if(!_impl->filters_to_ignore.empty())
{
observe("s|" + r + ":" + roc_identifier + ":SEL", fail);
}
observe("s|" + r + ":" + roc_identifier + ":ALL", false);
}
}
}
void bcf_remove_alleles(const bcf_hdr_t *header, bcf1_t *line, int rm_mask)
{
int *map = (int*) calloc(line->n_allele, sizeof(int));
// create map of indexes from old to new ALT numbering and modify ALT
kstring_t str = {0,0,0};
kputs(line->d.allele[0], &str);
int nrm = 0, i,j; // i: ori alleles, j: new alleles
for (i=1, j=1; i<line->n_allele; i++)
{
if ( rm_mask & 1<<i )
{
// remove this allele
line->d.allele[i] = NULL;
nrm++;
continue;
}
kputc(',', &str);
kputs(line->d.allele[i], &str);
map[i] = j;
j++;
}
if ( !nrm ) { free(map); free(str.s); return; }
int nR_ori = line->n_allele;
int nR_new = line->n_allele-nrm;
assert(nR_new > 0); // should not be able to remove reference allele
int nA_ori = nR_ori-1;
int nA_new = nR_new-1;
int nG_ori = nR_ori*(nR_ori + 1)/2;
int nG_new = nR_new*(nR_new + 1)/2;
bcf_update_alleles_str(header, line, str.s);
// remove from Number=G, Number=R and Number=A INFO fields.
uint8_t *dat = NULL;
int mdat = 0, ndat = 0, mdat_bytes = 0, nret;
for (i=0; i<line->n_info; i++)
{
bcf_info_t *info = &line->d.info[i];
int vlen = bcf_hdr_id2length(header,BCF_HL_INFO,info->key);
if ( vlen!=BCF_VL_A && vlen!=BCF_VL_G && vlen!=BCF_VL_R ) continue; // no need to change
int type = bcf_hdr_id2type(header,BCF_HL_INFO,info->key);
if ( type==BCF_HT_FLAG ) continue;
int size = 1;
if ( type==BCF_HT_REAL || type==BCF_HT_INT ) size = 4;
mdat = mdat_bytes / size;
nret = bcf_get_info_values(header, line, bcf_hdr_int2id(header,BCF_DT_ID,info->key), (void**)&dat, &mdat, type);
mdat_bytes = mdat * size;
if ( nret<0 )
{
fprintf(stderr,"[%s:%d %s] Could not access INFO/%s at %s:%d [%d]\n", __FILE__,__LINE__,__FUNCTION__,
bcf_hdr_int2id(header,BCF_DT_ID,info->key), bcf_seqname(header,line), line->pos+1, nret);
exit(1);
}
if ( type==BCF_HT_STR )
{
str.l = 0;
char *ss = (char*) dat, *se = (char*) dat;
if ( vlen==BCF_VL_A || vlen==BCF_VL_R )
{
int nexp, inc = 0;
if ( vlen==BCF_VL_A )
{
nexp = nA_ori;
inc = 1;
}
else
nexp = nR_ori;
for (j=0; j<nexp; j++)
{
if ( !*se ) break;
while ( *se && *se!=',' ) se++;
if ( rm_mask & 1<<(j+inc) )
{
if ( *se ) se++;
ss = se;
continue;
}
if ( str.l ) kputc(',',&str);
kputsn(ss,se-ss,&str);
if ( *se ) se++;
ss = se;
}
assert( j==nexp );
}
else // Number=G, assuming diploid genotype
{
int k = 0, n = 0;
for (j=0; j<nR_ori; j++)
{
for (k=0; k<=j; k++)
{
if ( !*se ) break;
while ( *se && *se!=',' ) se++;
n++;
if ( rm_mask & 1<<j || rm_mask & 1<<k )
{
if ( *se ) se++;
ss = se;
continue;
}
if ( str.l ) kputc(',',&str);
kputsn(ss,se-ss,&str);
if ( *se ) se++;
ss = se;
}
if ( !*se ) break;
}
assert( n=nG_ori );
}
nret = bcf_update_info(header, line, bcf_hdr_int2id(header,BCF_DT_ID,info->key), (void*)str.s, str.l, type);
if ( nret<0 )
{
fprintf(stderr,"[%s:%d %s] Could not update INFO/%s at %s:%d [%d]\n", __FILE__,__LINE__,__FUNCTION__,
bcf_hdr_int2id(header,BCF_DT_ID,info->key), bcf_seqname(header,line), line->pos+1, nret);
exit(1);
}
continue;
}
if ( vlen==BCF_VL_A || vlen==BCF_VL_R )
{
int inc = 0, ntop;
if ( vlen==BCF_VL_A )
{
assert( nret==nA_ori );
ntop = nA_ori;
ndat = nA_new;
inc = 1;
}
else
{
assert( nret==nR_ori );
ntop = nR_ori;
ndat = nR_new;
}
int k = 0;
#define BRANCH(type_t,is_vector_end) \
{ \
type_t *ptr = (type_t*) dat; \
int size = sizeof(type_t); \
for (j=0; j<ntop; j++) /* j:ori, k:new */ \
{ \
if ( is_vector_end ) { memcpy(dat+k*size, dat+j*size, size); break; } \
if ( rm_mask & 1<<(j+inc) ) continue; \
if ( j!=k ) memcpy(dat+k*size, dat+j*size, size); \
k++; \
} \
}
switch (type)
{
case BCF_HT_INT: BRANCH(int32_t,ptr[j]==bcf_int32_vector_end); break;
case BCF_HT_REAL: BRANCH(float,bcf_float_is_vector_end(ptr[j])); break;
}
#undef BRANCH
}
else // Number=G
{
assert( nret==nG_ori );
int k, l_ori = -1, l_new = 0;
ndat = nG_new;
#define BRANCH(type_t,is_vector_end) \
{ \
type_t *ptr = (type_t*) dat; \
int size = sizeof(type_t); \
for (j=0; j<nR_ori; j++) \
{ \
for (k=0; k<=j; k++) \
{ \
l_ori++; \
if ( is_vector_end ) { memcpy(dat+l_new*size, dat+l_ori*size, size); break; } \
if ( rm_mask & 1<<j || rm_mask & 1<<k ) continue; \
if ( l_ori!=l_new ) memcpy(dat+l_new*size, dat+l_ori*size, size); \
l_new++; \
} \
} \
}
switch (type)
{
case BCF_HT_INT: BRANCH(int32_t,ptr[l_ori]==bcf_int32_vector_end); break;
case BCF_HT_REAL: BRANCH(float,bcf_float_is_vector_end(ptr[l_ori])); break;
}
#undef BRANCH
}
nret = bcf_update_info(header, line, bcf_hdr_int2id(header,BCF_DT_ID,info->key), (void*)dat, ndat, type);
if ( nret<0 )
{
fprintf(stderr,"[%s:%d %s] Could not update INFO/%s at %s:%d [%d]\n", __FILE__,__LINE__,__FUNCTION__,
bcf_hdr_int2id(header,BCF_DT_ID,info->key), bcf_seqname(header,line), line->pos+1, nret);
exit(1);
}
}
// Update GT fields, the allele indexes might have changed
for (i=1; i<line->n_allele; i++) if ( map[i]!=i ) break;
if ( i<line->n_allele )
{
mdat = mdat_bytes / 4; // sizeof(int32_t)
nret = bcf_get_genotypes(header,line,(void**)&dat,&mdat);
mdat_bytes = mdat * 4;
if ( nret>0 )
{
nret /= line->n_sample;
int32_t *ptr = (int32_t*) dat;
for (i=0; i<line->n_sample; i++)
{
for (j=0; j<nret; j++)
{
if ( ptr[j]==bcf_gt_missing ) continue;
if ( ptr[j]==bcf_int32_vector_end ) break;
int al = bcf_gt_allele(ptr[j]);
assert( al<nR_ori && map[al]>=0 );
ptr[j] = (map[al]+1)<<1 | (ptr[j]&1);
}
ptr += nret;
}
bcf_update_genotypes(header, line, (void*)dat, nret*line->n_sample);
}
}
// Remove from Number=G, Number=R and Number=A FORMAT fields.
// Assuming haploid or diploid GTs
for (i=0; i<line->n_fmt; i++)
{
bcf_fmt_t *fmt = &line->d.fmt[i];
int vlen = bcf_hdr_id2length(header,BCF_HL_FMT,fmt->id);
if ( vlen!=BCF_VL_A && vlen!=BCF_VL_G && vlen!=BCF_VL_R ) continue; // no need to change
int type = bcf_hdr_id2type(header,BCF_HL_FMT,fmt->id);
if ( type==BCF_HT_FLAG ) continue;
int size = 1;
if ( type==BCF_HT_REAL || type==BCF_HT_INT ) size = 4;
mdat = mdat_bytes / size;
nret = bcf_get_format_values(header, line, bcf_hdr_int2id(header,BCF_DT_ID,fmt->id), (void**)&dat, &mdat, type);
mdat_bytes = mdat * size;
if ( nret<0 )
{
fprintf(stderr,"[%s:%d %s] Could not access FORMAT/%s at %s:%d [%d]\n", __FILE__,__LINE__,__FUNCTION__,
bcf_hdr_int2id(header,BCF_DT_ID,fmt->id), bcf_seqname(header,line), line->pos+1, nret);
exit(1);
}
if ( type==BCF_HT_STR )
{
int size = nret/line->n_sample; // number of bytes per sample
str.l = 0;
if ( vlen==BCF_VL_A || vlen==BCF_VL_R )
{
int nexp, inc = 0;
if ( vlen==BCF_VL_A )
{
nexp = nA_ori;
inc = 1;
}
else
nexp = nR_ori;
for (j=0; j<line->n_sample; j++)
{
char *ss = ((char*)dat) + j*size, *se = ss + size, *ptr = ss;
int k_src = 0, k_dst = 0, l = str.l;
for (k_src=0; k_src<nexp; k_src++)
{
if ( ptr>=se || !*ptr) break;
while ( ptr<se && *ptr && *ptr!=',' ) ptr++;
if ( rm_mask & 1<<(k_src+inc) )
{
ss = ++ptr;
continue;
}
if ( k_dst ) kputc(',',&str);
kputsn(ss,ptr-ss,&str);
ss = ++ptr;
k_dst++;
}
assert( k_src==nexp );
l = str.l - l;
for (; l<size; l++) kputc(0, &str);
}
}
else // Number=G, diploid or haploid
{
for (j=0; j<line->n_sample; j++)
{
char *ss = ((char*)dat) + j*size, *se = ss + size, *ptr = ss;
int k_src = 0, k_dst = 0, l = str.l;
int nexp = 0; // diploid or haploid?
while ( ptr<se )
{
if ( !*ptr ) break;
if ( *ptr==',' ) nexp++;
ptr++;
}
if ( ptr!=ss ) nexp++;
assert( nexp==nG_ori || nexp==nR_ori );
ptr = ss;
if ( nexp==nG_ori ) // diploid
{
int ia, ib;
for (ia=0; ia<nR_ori; ia++)
{
for (ib=0; ib<=ia; ib++)
{
if ( ptr>=se || !*ptr ) break;
while ( ptr<se && *ptr && *ptr!=',' ) ptr++;
if ( rm_mask & 1<<ia || rm_mask & 1<<ib )
{
ss = ++ptr;
continue;
}
if ( k_dst ) kputc(',',&str);
kputsn(ss,ptr-ss,&str);
ss = ++ptr;
k_dst++;
}
if ( ptr>=se || !*ptr ) break;
}
}
else // haploid
{
for (k_src=0; k_src<nR_ori; k_src++)
{
if ( ptr>=se || !*ptr ) break;
while ( ptr<se && *ptr && *ptr!=',' ) ptr++;
if ( rm_mask & 1<<k_src )
{
ss = ++ptr;
continue;
}
if ( k_dst ) kputc(',',&str);
kputsn(ss,ptr-ss,&str);
ss = ++ptr;
k_dst++;
}
assert( k_src==nR_ori );
l = str.l - l;
for (; l<size; l++) kputc(0, &str);
}
}
}
nret = bcf_update_format(header, line, bcf_hdr_int2id(header,BCF_DT_ID,fmt->id), (void*)str.s, str.l, type);
if ( nret<0 )
{
fprintf(stderr,"[%s:%d %s] Could not update FORMAT/%s at %s:%d [%d]\n", __FILE__,__LINE__,__FUNCTION__,
bcf_hdr_int2id(header,BCF_DT_ID,fmt->id), bcf_seqname(header,line), line->pos+1, nret);
exit(1);
}
continue;
}
int nori = nret / line->n_sample;
if ( vlen==BCF_VL_A || vlen==BCF_VL_R || (vlen==BCF_VL_G && nori==nR_ori) ) // Number=A, R or haploid Number=G
{
int ntop, inc = 0;
if ( vlen==BCF_VL_A )
{
assert( nori==nA_ori ); // todo: will fail if all values are missing
ntop = nA_ori;
ndat = nA_new*line->n_sample;
inc = 1;
}
else
{
assert( nori==nR_ori ); // todo: will fail if all values are missing
ntop = nR_ori;
ndat = nR_new*line->n_sample;
}
#define BRANCH(type_t,is_vector_end) \
{ \
for (j=0; j<line->n_sample; j++) \
{ \
type_t *ptr_src = ((type_t*)dat) + j*nori; \
type_t *ptr_dst = ((type_t*)dat) + j*nA_new; \
int size = sizeof(type_t); \
int k_src, k_dst = 0; \
for (k_src=0; k_src<ntop; k_src++) \
{ \
if ( is_vector_end ) { memcpy(ptr_dst+k_dst, ptr_src+k_src, size); break; } \
if ( rm_mask & 1<<(k_src+inc) ) continue; \
if ( k_src!=k_dst ) memcpy(ptr_dst+k_dst, ptr_src+k_src, size); \
k_dst++; \
} \
} \
}
switch (type)
{
case BCF_HT_INT: BRANCH(int32_t,ptr_src[k_src]==bcf_int32_vector_end); break;
case BCF_HT_REAL: BRANCH(float,bcf_float_is_vector_end(ptr_src[k_src])); break;
}
#undef BRANCH
}
else // Number=G, diploid or mixture of haploid+diploid
{
assert( nori==nG_ori );
ndat = nG_new*line->n_sample;
#define BRANCH(type_t,is_vector_end) \
{ \
for (j=0; j<line->n_sample; j++) \
{ \
type_t *ptr_src = ((type_t*)dat) + j*nori; \
type_t *ptr_dst = ((type_t*)dat) + j*nG_new; \
int size = sizeof(type_t); \
int ia, ib, k_dst = 0, k_src; \
int nset = 0; /* haploid or diploid? */ \
for (k_src=0; k_src<nG_ori; k_src++) { if ( is_vector_end ) break; nset++; } \
if ( nset==nR_ori ) /* haploid */ \
{ \
for (k_src=0; k_src<nR_ori; k_src++) \
{ \
if ( rm_mask & 1<<k_src ) continue; \
if ( k_src!=k_dst ) memcpy(ptr_dst+k_dst, ptr_src+k_src, size); \
k_dst++; \
} \
memcpy(ptr_dst+k_dst, ptr_src+k_src, size); \
} \
else /* diploid */ \
{ \
k_src = -1; \
for (ia=0; ia<nR_ori; ia++) \
{ \
for (ib=0; ib<=ia; ib++) \
{ \
k_src++; \
if ( is_vector_end ) { memcpy(ptr_dst+k_dst, ptr_src+k_src, size); ia = nR_ori; break; } \
if ( rm_mask & 1<<ia || rm_mask & 1<<ib ) continue; \
if ( k_src!=k_dst ) memcpy(ptr_dst+k_dst, ptr_src+k_src, size); \
k_dst++; \
} \
} \
} \
} \
}
switch (type)
{
case BCF_HT_INT: BRANCH(int32_t,ptr_src[k_src]==bcf_int32_vector_end); break;
case BCF_HT_REAL: BRANCH(float,bcf_float_is_vector_end(ptr_src[k_src])); break;
}
#undef BRANCH
}
nret = bcf_update_format(header, line, bcf_hdr_int2id(header,BCF_DT_ID,fmt->id), (void*)dat, ndat, type);
if ( nret<0 )
{
fprintf(stderr,"[%s:%d %s] Could not update FORMAT/%s at %s:%d [%d]\n", __FILE__,__LINE__,__FUNCTION__,
bcf_hdr_int2id(header,BCF_DT_ID,fmt->id), bcf_seqname(header,line), line->pos+1, nret);
exit(1);
}
}
free(dat);
free(str.s);
free(map);
}
int run(int argc, char **argv)
{
char *trio_samples = NULL, *trio_file = NULL, *rules_fname = NULL, *rules_string = NULL;
memset(&args,0,sizeof(args_t));
args.mode = 0;
args.output_fname = "-";
static struct option loptions[] =
{
{"trio",1,0,'t'},
{"trio-file",1,0,'T'},
{"delete",0,0,'d'},
{"list",1,0,'l'},
{"count",0,0,'c'},
{"rules",1,0,'r'},
{"rules-file",1,0,'R'},
{"output",required_argument,NULL,'o'},
{"output-type",required_argument,NULL,'O'},
{0,0,0,0}
};
int c;
while ((c = getopt_long(argc, argv, "?ht:T:l:cdr:R:o:O:",loptions,NULL)) >= 0)
{
switch (c)
{
case 'o': args.output_fname = optarg; break;
case 'O':
switch (optarg[0]) {
case 'b': args.output_type = FT_BCF_GZ; break;
case 'u': args.output_type = FT_BCF; break;
case 'z': args.output_type = FT_VCF_GZ; break;
case 'v': args.output_type = FT_VCF; break;
default: error("The output type \"%s\" not recognised\n", optarg);
};
break;
case 'R': rules_fname = optarg; break;
case 'r': rules_string = optarg; break;
case 'd': args.mode |= MODE_DELETE; break;
case 'c': args.mode |= MODE_COUNT; break;
case 'l':
if ( !strcmp("+",optarg) ) args.mode |= MODE_LIST_GOOD;
else if ( !strcmp("x",optarg) ) args.mode |= MODE_LIST_BAD;
else error("The argument not recognised: --list %s\n", optarg);
break;
case 't': trio_samples = optarg; break;
case 'T': trio_file = optarg; break;
case 'h':
case '?':
default: error("%s",usage()); break;
}
}
if ( rules_fname )
args.rules = regidx_init(rules_fname, parse_rules, NULL, sizeof(rule_t), &args);
else
args.rules = init_rules(&args, rules_string);
if ( !args.rules ) return -1;
args.itr = regitr_init(args.rules);
args.itr_ori = regitr_init(args.rules);
char *fname = NULL;
if ( optind>=argc || argv[optind][0]=='-' )
{
if ( !isatty(fileno((FILE *)stdin)) ) fname = "-"; // reading from stdin
else error("%s",usage());
}
else
fname = argv[optind];
if ( !trio_samples && !trio_file ) error("Expected the -t/T option\n");
if ( !args.mode ) error("Expected one of the -c, -d or -l options\n");
if ( args.mode&MODE_DELETE && !(args.mode&(MODE_LIST_GOOD|MODE_LIST_BAD)) ) args.mode |= MODE_LIST_GOOD|MODE_LIST_BAD;
args.sr = bcf_sr_init();
if ( !bcf_sr_add_reader(args.sr, fname) ) error("Failed to open %s: %s\n", fname,bcf_sr_strerror(args.sr->errnum));
args.hdr = bcf_sr_get_header(args.sr, 0);
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));
bcf_hdr_write(args.out_fh, args.hdr);
int i, n = 0;
char **list;
if ( trio_samples )
{
args.ntrios = 1;
args.trios = (trio_t*) calloc(1,sizeof(trio_t));
list = hts_readlist(trio_samples, 0, &n);
if ( n!=3 ) error("Expected three sample names with -t\n");
args.trios[0].imother = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, list[0]);
args.trios[0].ifather = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, list[1]);
args.trios[0].ichild = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, list[2]);
for (i=0; i<n; i++) free(list[i]);
free(list);
}
if ( trio_file )
{
list = hts_readlist(trio_file, 1, &n);
args.ntrios = n;
args.trios = (trio_t*) calloc(n,sizeof(trio_t));
for (i=0; i<n; i++)
{
char *ss = list[i], *se;
se = strchr(ss, ',');
if ( !se ) error("Could not parse %s: %s\n",trio_file, ss);
*se = 0;
args.trios[i].imother = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, ss);
if ( args.trios[i].imother<0 ) error("No such sample: \"%s\"\n", ss);
ss = ++se;
se = strchr(ss, ',');
if ( !se ) error("Could not parse %s\n",trio_file);
*se = 0;
args.trios[i].ifather = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, ss);
if ( args.trios[i].ifather<0 ) error("No such sample: \"%s\"\n", ss);
ss = ++se;
if ( *ss=='\0' ) error("Could not parse %s\n",trio_file);
args.trios[i].ichild = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, ss);
if ( args.trios[i].ichild<0 ) error("No such sample: \"%s\"\n", ss);
free(list[i]);
}
free(list);
}
while ( bcf_sr_next_line(args.sr) )
{
bcf1_t *line = bcf_sr_get_line(args.sr,0);
line = process(line);
if ( line )
{
if ( line->errcode ) error("TODO: Unchecked error (%d), exiting\n",line->errcode);
bcf_write1(args.out_fh, args.hdr, line);
}
}
fprintf(stderr,"# [1]nOK\t[2]nBad\t[3]nSkipped\t[4]Trio\n");
for (i=0; i<args.ntrios; i++)
{
trio_t *trio = &args.trios[i];
fprintf(stderr,"%d\t%d\t%d\t%s,%s,%s\n",
trio->nok,trio->nbad,args.nrec-(trio->nok+trio->nbad),
bcf_hdr_int2id(args.hdr, BCF_DT_SAMPLE, trio->imother),
bcf_hdr_int2id(args.hdr, BCF_DT_SAMPLE, trio->ifather),
bcf_hdr_int2id(args.hdr, BCF_DT_SAMPLE, trio->ichild)
);
}
free(args.gt_arr);
free(args.trios);
regitr_destroy(args.itr);
regitr_destroy(args.itr_ori);
regidx_destroy(args.rules);
bcf_sr_destroy(args.sr);
if ( hts_close(args.out_fh)!=0 ) error("Error: close failed\n");
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);
}
static void init_data(args_t *args)
{
int i;
args->hdr = args->files->readers[0].header;
if (args->calc_ac && args->update_info)
{
bcf_hdr_append(args->hdr,"##INFO=<ID=AC,Number=A,Type=Integer,Description=\"Allele count in genotypes\">");
bcf_hdr_append(args->hdr,"##INFO=<ID=AN,Number=1,Type=Integer,Description=\"Total number of alleles in called genotypes\">");
}
bcf_hdr_append_version(args->hdr, args->argc, args->argv, "bcftools_view");
// setup sample data
if (args->sample_names)
{
void *hdr_samples = khash_str2int_init();
for (i=0; i<bcf_hdr_nsamples(args->hdr); i++)
khash_str2int_inc(hdr_samples, bcf_hdr_int2id(args->hdr,BCF_DT_SAMPLE,i));
void *exclude = (args->sample_names[0]=='^') ? khash_str2int_init() : NULL;
int nsmpl;
char **smpl = NULL;
args->samples = NULL; args->n_samples = 0;
smpl = hts_readlist(exclude ? &args->sample_names[1] : args->sample_names, args->sample_is_file, &nsmpl);
if ( !smpl )
{
error("Could not read the list: \"%s\"\n", exclude ? &args->sample_names[1] : args->sample_names);
}
if ( exclude )
{
for (i=0; i<nsmpl; i++) {
if (!khash_str2int_has_key(hdr_samples,smpl[i])) {
if (args->force_samples) {
fprintf(stderr, "Warn: exclude called for sample that does not exist in header: \"%s\"... skipping\n", smpl[i]);
} else {
error("Error: exclude called for sample that does not exist in header: \"%s\". Use \"--force-samples\" to ignore this error.\n", smpl[i]);
}
}
khash_str2int_inc(exclude, smpl[i]);
}
for (i=0; i<bcf_hdr_nsamples(args->hdr); i++)
{
if ( exclude && khash_str2int_has_key(exclude,bcf_hdr_int2id(args->hdr,BCF_DT_SAMPLE,i)) ) continue;
args->samples = (char**) realloc(args->samples, (args->n_samples+1)*sizeof(const char*));
args->samples[args->n_samples++] = strdup(bcf_hdr_int2id(args->hdr,BCF_DT_SAMPLE,i));
}
khash_str2int_destroy(exclude);
}
else
{
for (i=0; i<nsmpl; i++) {
if (!khash_str2int_has_key(hdr_samples,smpl[i])) {
if (args->force_samples) {
fprintf(stderr, "Warn: subset called for sample that does not exist in header: \"%s\"... skipping\n", smpl[i]);
continue;
} else {
error("Error: subset called for sample that does not exist in header: \"%s\". Use \"--force-samples\" to ignore this error.\n", smpl[i]);
}
}
args->samples = (char**) realloc(args->samples, (args->n_samples+1)*sizeof(const char*));
args->samples[args->n_samples++] = strdup(smpl[i]);
}
}
for (i=0; i<nsmpl; i++) free(smpl[i]);
free(smpl);
khash_str2int_destroy(hdr_samples);
if (args->n_samples == 0) {
fprintf(stderr, "Warn: subsetting has removed all samples\n");
args->sites_only = 1;
}
}
if (args->n_samples)
args->imap = (int*)malloc(args->n_samples * sizeof(int));
// determine variant types to include/exclude
if (args->include_types || args->exclude_types) {
if (args->include_types && args->exclude_types) {
fprintf(stderr, "Error: only supply one of --include-types, --exclude-types options\n");
exit(1);
}
char **type_list = 0;
int m = 0, n = 0;
const char *q, *p;
for (q = p = args->include_types ? args->include_types : args->exclude_types;; ++p) {
if (*p == ',' || *p == 0) {
if (m == n) {
m = m? m<<1 : 16;
type_list = (char**)realloc(type_list, m * sizeof(char*));
}
type_list[n] = (char*)calloc(p - q + 1, 1);
strncpy(type_list[n++], q, p - q);
q = p + 1;
if (*p == 0) break;
}
}
type_list = (char**)realloc(type_list, n * sizeof(char*));
if (args->include_types) {
args->include = 0;
for (i = 0; i < n; ++i) {
if (strcmp(type_list[i], "snps") == 0) args->include |= VCF_SNP;
else if (strcmp(type_list[i], "indels") == 0) args->include |= VCF_INDEL;
else if (strcmp(type_list[i], "mnps") == 0) args->include |= VCF_MNP;
else if (strcmp(type_list[i], "other") == 0) args->include |= VCF_OTHER;
else {
fprintf(stderr, "[E::%s] unknown type\n", type_list[i]);
fprintf(stderr, "Accepted types are snps, indels, mnps, other\n");
exit(1);
}
}
}
if (args->exclude_types) {
args->exclude = 0;
for (i = 0; i < n; ++i) {
if (strcmp(type_list[i], "snps") == 0) args->exclude |= VCF_SNP;
else if (strcmp(type_list[i], "indels") == 0) args->exclude |= VCF_INDEL;
else if (strcmp(type_list[i], "mnps") == 0) args->exclude |= VCF_MNP;
else if (strcmp(type_list[i], "other") == 0) args->exclude |= VCF_OTHER;
else {
fprintf(stderr, "[E::%s] unknown type\n", type_list[i]);
fprintf(stderr, "Accepted types are snps, indels, mnps, other\n");
exit(1);
}
}
}
for (i = 0; i < n; ++i)
free(type_list[i]);
free(type_list);
}
// setup output
char modew[8];
strcpy(modew, "w");
if (args->clevel >= 0 && args->clevel <= 9) sprintf(modew + 1, "%d", args->clevel);
if (args->output_type==FT_BCF) strcat(modew, "bu"); // uncompressed BCF
else if (args->output_type & FT_BCF) strcat(modew, "b"); // compressed BCF
else if (args->output_type & FT_GZ) strcat(modew,"z"); // compressed VCF
args->out = hts_open(args->fn_out ? args->fn_out : "-", modew);
if ( !args->out ) error("%s: %s\n", args->fn_out,strerror(errno));
// headers: hdr=full header, hsub=subset header, hnull=sites only header
if (args->sites_only){
args->hnull = bcf_hdr_subset(args->hdr, 0, 0, 0);
bcf_hdr_remove(args->hnull, BCF_HL_FMT, NULL);
}
if (args->n_samples > 0)
{
args->hsub = bcf_hdr_subset(args->hdr, args->n_samples, args->samples, args->imap);
if ( !args->hsub ) error("Error occurred while subsetting samples\n");
if ( args->n_samples != bcf_hdr_nsamples(args->hsub) )
{
int i;
for (i=0; i<args->n_samples; i++)
if ( args->imap[i]<0 ) error("Error: No such sample: \"%s\"\n", args->samples[i]);
}
}
if ( args->filter_str )
args->filter = filter_init(args->hdr, args->filter_str);
}
void flush_viterbi(args_t *args)
{
const char *s1, *s2, *s3 = NULL;
if ( args->mode==C_UNRL )
{
s1 = bcf_hdr_int2id(args->hdr,BCF_DT_SAMPLE,args->isample);
s2 = bcf_hdr_int2id(args->hdr,BCF_DT_SAMPLE,args->jsample);
}
else if ( args->mode==C_TRIO )
{
s1 = bcf_hdr_int2id(args->hdr,BCF_DT_SAMPLE,args->imother);
s3 = bcf_hdr_int2id(args->hdr,BCF_DT_SAMPLE,args->ifather);
s2 = bcf_hdr_int2id(args->hdr,BCF_DT_SAMPLE,args->ichild);
}
if ( !args->fp )
{
kstring_t str = {0,0,0};
kputs(args->prefix, &str);
kputs(".dat", &str);
args->fp = fopen(str.s,"w");
if ( !args->fp ) error("%s: %s\n", str.s,strerror(errno));
free(str.s);
fprintf(args->fp,"# SG, shared segment\t[2]Chromosome\t[3]Start\t[4]End\t[5]%s:1\t[6]%s:2\n",s2,s2);
fprintf(args->fp,"# SW, number of switches\t[3]Sample\t[4]Chromosome\t[5]nHets\t[5]nSwitches\t[6]switch rate\n");
}
hmm_run_viterbi(args->hmm,args->nsites,args->eprob,args->sites);
uint8_t *vpath = hmm_get_viterbi_path(args->hmm);
int i, iprev = -1, prev_state = -1, nstates = hmm_get_nstates(args->hmm);
int nswitch_mother = 0, nswitch_father = 0;
for (i=0; i<args->nsites; i++)
{
int state = vpath[i*nstates];
if ( state!=prev_state || i+1==args->nsites )
{
uint32_t start = iprev>=0 ? args->sites[iprev]+1 : 1, end = i>0 ? args->sites[i-1] : 1;
const char *chr = bcf_hdr_id2name(args->hdr,args->prev_rid);
if ( args->mode==C_UNRL )
{
switch (prev_state)
{
case UNRL_0x0x:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:1\t-\n", chr,start,end,s1); break;
case UNRL_0xx0:
fprintf(args->fp,"SG\t%s\t%d\t%d\t-\t%s:1\n", chr,start,end,s1); break;
case UNRL_x00x:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:2\t-\n", chr,start,end,s1); break;
case UNRL_x0x0:
fprintf(args->fp,"SG\t%s\t%d\t%d\t-\t%s:2\n", chr,start,end,s1); break;
case UNRL_0101:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:1\t%s:2\n", chr,start,end,s1,s1); break;
case UNRL_0110:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:2\t%s:1\n", chr,start,end,s1,s1); break;
}
}
else if ( args->mode==C_TRIO )
{
switch (prev_state)
{
case TRIO_AC:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:1\t%s:1\n", chr,start,end,s1,s3); break;
case TRIO_AD:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:1\t%s:2\n", chr,start,end,s1,s3); break;
case TRIO_BC:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:2\t%s:1\n", chr,start,end,s1,s3); break;
case TRIO_BD:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:2\t%s:2\n", chr,start,end,s1,s3); break;
case TRIO_CA:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:1\t%s:1\n", chr,start,end,s3,s1); break;
case TRIO_DA:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:2\t%s:1\n", chr,start,end,s3,s1); break;
case TRIO_CB:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:1\t%s:2\n", chr,start,end,s3,s1); break;
case TRIO_DB:
fprintf(args->fp,"SG\t%s\t%d\t%d\t%s:2\t%s:2\n", chr,start,end,s3,s1); break;
}
if ( hap_switch[state][prev_state] & SW_MOTHER ) nswitch_mother++;
if ( hap_switch[state][prev_state] & SW_FATHER ) nswitch_father++;
}
iprev = i-1;
}
prev_state = state;
}
float mrate = args->nhet_mother>1 ? (float)nswitch_mother/(args->nhet_mother-1) : 0;
float frate = args->nhet_father>1 ? (float)nswitch_father/(args->nhet_father-1) : 0;
fprintf(args->fp,"SW\t%s\t%s\t%d\t%d\t%f\n", s1,bcf_hdr_id2name(args->hdr,args->prev_rid),args->nhet_mother,nswitch_mother,mrate);
fprintf(args->fp,"SW\t%s\t%s\t%d\t%d\t%f\n", s3,bcf_hdr_id2name(args->hdr,args->prev_rid),args->nhet_father,nswitch_father,frate);
args->nsites = 0;
args->nhet_father = args->nhet_mother = 0;
}