int parse_format_flag(const char *str)
{
int i, flag = 0, n_tags;
char **tags = hts_readlist(str, 0, &n_tags);
for(i=0; i<n_tags; i++)
{
if ( !strcasecmp(tags[i],"DP") || !strcasecmp(tags[i],"FORMAT/DP") || !strcasecmp(tags[i],"FMT/DP") ) flag |= B2B_FMT_DP;
else if ( !strcasecmp(tags[i],"DV") || !strcasecmp(tags[i],"FORMAT/DV") || !strcasecmp(tags[i],"FMT/DV") ) { flag |= B2B_FMT_DV; fprintf(stderr, "[warning] tag DV functional, but deprecated. Please switch to `AD` in future.\n"); }
else if ( !strcasecmp(tags[i],"SP") || !strcasecmp(tags[i],"FORMAT/SP") || !strcasecmp(tags[i],"FMT/SP") ) flag |= B2B_FMT_SP;
else if ( !strcasecmp(tags[i],"DP4") || !strcasecmp(tags[i],"FORMAT/DP4") || !strcasecmp(tags[i],"FMT/DP4") ) { flag |= B2B_FMT_DP4; fprintf(stderr, "[warning] tag DP4 functional, but deprecated. Please switch to `ADF` and `ADR` in future.\n"); }
else if ( !strcasecmp(tags[i],"DPR") || !strcasecmp(tags[i],"FORMAT/DPR") || !strcasecmp(tags[i],"FMT/DPR") ) { flag |= B2B_FMT_DPR; fprintf(stderr, "[warning] tag DPR functional, but deprecated. Please switch to `AD` in future.\n"); }
else if ( !strcasecmp(tags[i],"INFO/DPR") ) { flag |= B2B_INFO_DPR; fprintf(stderr, "[warning] tag INFO/DPR functional, but deprecated. Please switch to `INFO/AD` in future.\n"); }
else if ( !strcasecmp(tags[i],"AD") || !strcasecmp(tags[i],"FORMAT/AD") || !strcasecmp(tags[i],"FMT/AD") ) flag |= B2B_FMT_AD;
else if ( !strcasecmp(tags[i],"ADF") || !strcasecmp(tags[i],"FORMAT/ADF") || !strcasecmp(tags[i],"FMT/ADF") ) flag |= B2B_FMT_ADF;
else if ( !strcasecmp(tags[i],"ADR") || !strcasecmp(tags[i],"FORMAT/ADR") || !strcasecmp(tags[i],"FMT/ADR") ) flag |= B2B_FMT_ADR;
else if ( !strcasecmp(tags[i],"INFO/AD") ) flag |= B2B_INFO_AD;
else if ( !strcasecmp(tags[i],"INFO/ADF") ) flag |= B2B_INFO_ADF;
else if ( !strcasecmp(tags[i],"INFO/ADR") ) flag |= B2B_INFO_ADR;
else
{
fprintf(stderr,"Could not parse tag \"%s\" in \"%s\"\n", tags[i], str);
exit(EXIT_FAILURE);
}
free(tags[i]);
}
if (n_tags) free(tags);
return flag;
}
int bam_smpl_add_samples(bam_smpl_t *bsmpl, char *list, int is_file)
{
if ( list[0]!='^' ) bsmpl->sample_logic = 1;
else list++;
int i, nsamples = 0;
char **samples = hts_readlist(list, is_file, &nsamples);
if ( !nsamples ) return 0;
kstring_t ori = {0,0,0};
kstring_t ren = {0,0,0};
bsmpl->sample_list = khash_str2str_init();
for (i=0; i<nsamples; i++)
{
char *ptr = samples[i];
ori.l = ren.l = 0;
int escaped = 0;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &ori);
escaped = 0;
ptr++;
}
if ( *ptr )
{
while ( *ptr && isspace(*ptr) ) ptr++;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &ren);
escaped = 0;
ptr++;
}
}
khash_str2str_set(bsmpl->sample_list,strdup(ori.s),strdup(ren.l?ren.s:ori.s));
free(samples[i]);
}
free(samples);
free(ori.s);
free(ren.s);
return nsamples;
}
static void init_data(args_t *args)
{
args->header = args->files->readers[0].header;
int i, nsamples = 0, *samples = NULL;
if ( args->sample_list && strcmp("-",args->sample_list) )
{
for (i=0; i<args->files->nreaders; i++)
{
int ret = bcf_hdr_set_samples(args->files->readers[i].header,args->sample_list,args->sample_is_file);
if ( ret<0 ) error("Error parsing the sample list\n");
else if ( ret>0 ) error("Sample name mismatch: sample #%d not found in the header\n", ret);
}
if ( args->sample_list[0]!='^' )
{
// the sample ordering may be different if not negated
int n;
char **smpls = hts_readlist(args->sample_list, args->sample_is_file, &n);
if ( !smpls ) error("Could not parse %s\n", args->sample_list);
if ( n!=bcf_hdr_nsamples(args->files->readers[0].header) )
error("The number of samples does not match, perhaps some are present multiple times?\n");
nsamples = bcf_hdr_nsamples(args->files->readers[0].header);
samples = (int*) malloc(sizeof(int)*nsamples);
for (i=0; i<n; i++)
{
samples[i] = bcf_hdr_id2int(args->files->readers[0].header, BCF_DT_SAMPLE,smpls[i]);
free(smpls[i]);
}
free(smpls);
}
}
args->convert = convert_init(args->header, samples, nsamples, args->format_str);
if ( args->allow_undef_tags ) convert_set_option(args->convert, allow_undef_tags, 1);
free(samples);
int max_unpack = convert_max_unpack(args->convert);
if ( args->filter_str )
{
args->filter = filter_init(args->header, args->filter_str);
max_unpack |= filter_max_unpack(args->filter);
}
args->files->max_unpack = max_unpack;
}
int parse_format_flag(const char *str)
{
int i, flag = 0, n_tags;
char **tags = hts_readlist(str, 0, &n_tags);
for(i=0; i<n_tags; i++)
{
if ( !strcasecmp(tags[i],"DP") ) flag |= B2B_FMT_DP;
else if ( !strcasecmp(tags[i],"DV") ) flag |= B2B_FMT_DV;
else if ( !strcasecmp(tags[i],"SP") ) flag |= B2B_FMT_SP;
else if ( !strcasecmp(tags[i],"DP4") ) flag |= B2B_FMT_DP4;
else if ( !strcasecmp(tags[i],"DPR") ) flag |= B2B_FMT_DPR;
else if ( !strcasecmp(tags[i],"INFO/DPR") ) flag |= B2B_INFO_DPR;
else
{
fprintf(stderr,"Could not parse tag \"%s\" in \"%s\"\n", tags[i], str);
exit(1);
}
free(tags[i]);
}
if (n_tags) free(tags);
return flag;
}
int parse_tags(args_t *args, const char *str)
{
int i, flag = 0, n_tags;
char **tags = hts_readlist(str, 0, &n_tags);
for(i=0; i<n_tags; i++)
{
if ( !strcasecmp(tags[i],"AN") ) flag |= SET_AN;
else if ( !strcasecmp(tags[i],"AC") ) flag |= SET_AC;
else if ( !strcasecmp(tags[i],"NS") ) flag |= SET_NS;
else if ( !strcasecmp(tags[i],"AC_Hom") ) flag |= SET_AC_Hom;
else if ( !strcasecmp(tags[i],"AC_Het") ) flag |= SET_AC_Het;
else if ( !strcasecmp(tags[i],"AC_Hemi") ) flag |= SET_AC_Hemi;
else if ( !strcasecmp(tags[i],"AF") ) flag |= SET_AF;
else
{
fprintf(stderr,"Error parsing \"--tags %s\": the tag \"%s\" is not supported\n", str,tags[i]);
exit(1);
}
free(tags[i]);
}
if (n_tags) free(tags);
return flag;
}
static void list_columns(args_t *args)
{
void *has_sample = NULL;
if ( args->sample_list )
{
has_sample = khash_str2int_init();
int i, nsmpl;
char **smpl = hts_readlist(args->sample_list, args->sample_is_file, &nsmpl);
for (i=0; i<nsmpl; i++) khash_str2int_inc(has_sample, smpl[i]);
free(smpl);
}
int i;
bcf_sr_t *reader = &args->files->readers[0];
for (i=0; i<bcf_hdr_nsamples(reader->header); i++)
{
if ( has_sample && !khash_str2int_has_key(has_sample, reader->header->samples[i]) ) continue;
printf("%s\n", reader->header->samples[i]);
}
if ( has_sample )
khash_str2int_destroy_free(has_sample);
}
int 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;
}
int bcf_sr_set_samples(bcf_srs_t *files, const char *fname, int is_file)
{
int i, j, nsmpl, free_smpl = 0;
char **smpl = NULL;
void *exclude = (fname[0]=='^') ? khash_str2int_init() : NULL;
if ( exclude || strcmp("-",fname) ) // "-" stands for all samples
{
smpl = hts_readlist(fname, is_file, &nsmpl);
if ( !smpl )
{
fprintf(stderr,"Could not read the file: \"%s\"\n", fname);
return 0;
}
if ( exclude )
{
for (i=0; i<nsmpl; i++)
khash_str2int_inc(exclude, smpl[i]);
}
free_smpl = 1;
}
if ( !smpl )
{
smpl = files->readers[0].header->samples; // intersection of all samples
nsmpl = bcf_hdr_nsamples(files->readers[0].header);
}
files->samples = NULL;
files->n_smpl = 0;
for (i=0; i<nsmpl; i++)
{
if ( exclude && khash_str2int_has_key(exclude,smpl[i]) ) continue;
int n_isec = 0;
for (j=0; j<files->nreaders; j++)
{
if ( bcf_hdr_id2int(files->readers[j].header, BCF_DT_SAMPLE, smpl[i])<0 ) break;
n_isec++;
}
if ( n_isec!=files->nreaders )
{
fprintf(stderr,"Warning: The sample \"%s\" was not found in %s, skipping\n", smpl[i], files->readers[n_isec].fname);
continue;
}
files->samples = (char**) realloc(files->samples, (files->n_smpl+1)*sizeof(const char*));
files->samples[files->n_smpl++] = strdup(smpl[i]);
}
if ( exclude ) khash_str2int_destroy(exclude);
if ( free_smpl )
{
for (i=0; i<nsmpl; i++) free(smpl[i]);
free(smpl);
}
if ( !files->n_smpl )
{
if ( files->nreaders>1 )
fprintf(stderr,"No samples in common.\n");
return 0;
}
for (i=0; i<files->nreaders; i++)
{
bcf_sr_t *reader = &files->readers[i];
reader->samples = (int*) malloc(sizeof(int)*files->n_smpl);
reader->n_smpl = files->n_smpl;
for (j=0; j<files->n_smpl; j++)
reader->samples[j] = bcf_hdr_id2int(reader->header, BCF_DT_SAMPLE, files->samples[j]);
}
return 1;
}
int init(int argc, char **argv, bcf_hdr_t *in, bcf_hdr_t *out)
{
memset(&args,0,sizeof(args_t));
int i;
static struct option loptions[] =
{
{"help", no_argument, 0,'h'},
{"sample-list", required_argument, 0,'s'},
{0,0,0,0}
};
char **smps_strs = NULL;
int c;
while ((c = getopt_long(argc, argv, "?s:h",loptions,NULL)) >= 0)
{
switch (c)
{
case 's': smps_strs = hts_readlist(optarg,0,&(args.n_sel_smps));
if ( args.n_sel_smps == 0 )
{
fprintf(stderr, "Sample specification not valid.\n");
error("%s", usage());
}
break;
case 'h': usage(); break;
case '?':
default: error("%s", usage()); break;
}
}
if ( optind != argc ) usage(); // too many files given
args.hdr = bcf_hdr_dup(in);
// Samples parsing from header and input option
if ( !bcf_hdr_nsamples(args.hdr) )
{
error("No samples in input file.\n");
}
args.nsmp = bcf_hdr_nsamples(args.hdr);
args.selected_smps = (int*) calloc(args.nsmp,sizeof(int));
for ( i = 0; i < args.n_sel_smps; i++ )
{
int ind = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, smps_strs[i]);
if ( ind == -1 )
{
error("Sample '%s' not in input vcf file.\n", smps_strs[i]);
} else {
args.selected_smps[ind] = 1;
}
free(smps_strs[i]);
}
free(smps_strs);
/*
fprintf(stderr, "Selected samples array:[");
for (i=0;i<args.nsmp;i++)
{
fprintf(stderr, " %i", args.selected_smps[i]);
}
fprintf(stderr, " ]\n");
*/
if ( bcf_hdr_id2int(args.hdr, BCF_DT_ID, "GT")<0 ) error("[E::%s] GT not present in the header\n", __func__);
args.gt_arr = NULL;
return 0;
}
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);
}
static void init_data(args_t *args)
{
args->prev_rid = args->skip_rid = -1;
args->hdr = args->files->readers[0].header;
if ( !args->sample )
{
if ( bcf_hdr_nsamples(args->hdr)>1 ) error("Missing the option -s, --sample\n");
args->sample = strdup(args->hdr->samples[0]);
}
if ( !bcf_hdr_nsamples(args->hdr) ) error("No samples in the VCF?\n");
// Set samples
kstring_t str = {0,0,0};
if ( args->estimate_AF && strcmp("-",args->estimate_AF) )
{
int i, n;
char **smpls = hts_readlist(args->estimate_AF, 1, &n);
// Make sure the query sample is included
for (i=0; i<n; i++)
if ( !strcmp(args->sample,smpls[i]) ) break;
// Add the query sample if not present
if ( i!=n ) kputs(args->sample, &str);
for (i=0; i<n; i++)
{
if ( str.l ) kputc(',', &str);
kputs(smpls[i], &str);
free(smpls[i]);
}
free(smpls);
}
else if ( !args->estimate_AF )
kputs(args->sample, &str);
if ( str.l )
{
int ret = bcf_hdr_set_samples(args->hdr, str.s, 0);
if ( ret<0 ) error("Error parsing the list of samples: %s\n", str.s);
else if ( ret>0 ) error("The %d-th sample not found in the VCF\n", ret);
}
if ( args->af_tag )
if ( !bcf_hdr_idinfo_exists(args->hdr,BCF_HL_INFO,bcf_hdr_id2int(args->hdr,BCF_DT_ID,args->af_tag)) )
error("No such INFO tag in the VCF: %s\n", args->af_tag);
args->nsmpl = bcf_hdr_nsamples(args->hdr);
args->ismpl = bcf_hdr_id2int(args->hdr, BCF_DT_SAMPLE, args->sample);
free(str.s);
int i;
for (i=0; i<256; i++) args->pl2p[i] = pow(10., -i/10.);
// Init transition matrix and HMM
double tprob[4];
MAT(tprob,2,STATE_HW,STATE_HW) = 1 - args->t2AZ;
MAT(tprob,2,STATE_HW,STATE_AZ) = args->t2HW;
MAT(tprob,2,STATE_AZ,STATE_HW) = args->t2AZ;
MAT(tprob,2,STATE_AZ,STATE_AZ) = 1 - args->t2HW;
if ( args->genmap_fname )
{
args->hmm = hmm_init(2, tprob, 0);
hmm_set_tprob_func(args->hmm, set_tprob_genmap, args);
}
else if ( args->rec_rate > 0 )
{
args->hmm = hmm_init(2, tprob, 0);
hmm_set_tprob_func(args->hmm, set_tprob_recrate, args);
}
else
args->hmm = hmm_init(2, tprob, 10000);
// print header
printf("# This file was produced by: bcftools roh(%s+htslib-%s)\n", bcftools_version(),hts_version());
printf("# The command line was:\tbcftools %s", args->argv[0]);
for (i=1; i<args->argc; i++)
printf(" %s",args->argv[i]);
printf("\n#\n");
printf("# [1]Chromosome\t[2]Position\t[3]State (0:HW, 1:AZ)\t[4]Quality\n");
}
int bam_smpl_add_readgroups(bam_smpl_t *bsmpl, char *list, int is_file)
{
if ( list[0]!='^' ) bsmpl->rg_logic = 1;
else list++;
int i, nrows = 0;
char **rows = hts_readlist(list, is_file, &nrows);
if ( !nrows ) return 0;
kstring_t fld1 = {0,0,0};
kstring_t fld2 = {0,0,0};
kstring_t fld3 = {0,0,0};
bsmpl->rg_list = khash_str2str_init();
for (i=0; i<nrows; i++)
{
char *ptr = rows[i];
fld1.l = fld2.l = fld3.l = 0;
int escaped = 0;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &fld1);
escaped = 0;
ptr++;
}
if ( *ptr )
{
while ( *ptr && isspace(*ptr) ) ptr++;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &fld2);
escaped = 0;
ptr++;
}
}
if ( *ptr )
{
while ( *ptr && isspace(*ptr) ) ptr++;
while ( *ptr )
{
if ( *ptr=='\\' && !escaped ) { escaped = 1; ptr++; continue; }
if ( isspace(*ptr) && !escaped ) break;
kputc(*ptr, &fld3);
escaped = 0;
ptr++;
}
}
if ( fld3.l )
{
// ID FILE SAMPLE
kputc('\t',&fld1);
kputs(fld2.s,&fld1);
fld2.l = 0;
kputs(fld3.s,&fld2);
}
// fld2.s now contains a new sample name. If NULL, use \t to keep the bam header name
char *value = khash_str2str_get(bsmpl->rg_list,fld1.s);
if ( !value )
khash_str2str_set(bsmpl->rg_list,strdup(fld1.s),strdup(fld2.l?fld2.s:"\t"));
else if ( strcmp(value,fld2.l?fld2.s:"\t") )
error("Error: The read group \"%s\" was assigned to two different samples: \"%s\" and \"%s\"\n", fld1.s,value,fld2.l?fld2.s:"\t");
free(rows[i]);
}
free(rows);
free(fld1.s);
free(fld2.s);
free(fld3.s);
return nrows;
}
int init(int argc, char **argv, bcf_hdr_t *in, bcf_hdr_t *out)
{
char *trio_samples = NULL, *unrelated_samples = NULL;
memset(&args,0,sizeof(args_t));
args.prev_rid = -1;
args.hdr = in;
args.pij = 2e-8;
args.pgt_err = 1e-9;
static struct option loptions[] =
{
{"prefix",1,0,'p'},
{"trio",1,0,'t'},
{"unrelated",1,0,'u'},
{0,0,0,0}
};
int c;
while ((c = getopt_long(argc, argv, "?ht:u:p:",loptions,NULL)) >= 0)
{
switch (c)
{
case 'p': args.prefix = optarg; break;
case 't': trio_samples = optarg; break;
case 'u': unrelated_samples = optarg; break;
case 'h':
case '?':
default: error("%s", usage()); break;
}
}
if ( optind != argc ) error(usage());
if ( trio_samples && unrelated_samples ) error("Expected only one of the -t/-u options\n");
if ( !trio_samples && !unrelated_samples ) error("Expected one of the -t/-u options\n");
if ( !args.prefix ) error("Expected the -p option\n");
int ret = bcf_hdr_set_samples(args.hdr, trio_samples ? trio_samples : unrelated_samples, 0);
if ( ret<0 ) error("Could not parse samples: %s\n", trio_samples ? trio_samples : unrelated_samples);
else if ( ret>0 ) error("%d-th sample not found: %s\n", ret,trio_samples ? trio_samples : unrelated_samples);
if ( trio_samples )
{
int i,n = 0;
char **list = hts_readlist(trio_samples, 0, &n);
if ( n!=3 ) error("Expected three sample names with -t\n");
args.imother = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, list[0]);
args.ifather = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, list[1]);
args.ichild = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, list[2]);
for (i=0; i<n; i++) free(list[i]);
free(list);
args.set_observed_prob = set_observed_prob_trio;
args.mode = C_TRIO;
init_hmm_trio(&args);
}
else
{
int i,n = 0;
char **list = hts_readlist(unrelated_samples, 0, &n);
if ( n!=2 ) error("Expected two sample names with -u\n");
args.isample = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, list[0]);
args.jsample = bcf_hdr_id2int(args.hdr, BCF_DT_SAMPLE, list[1]);
for (i=0; i<n; i++) free(list[i]);
free(list);
args.set_observed_prob = set_observed_prob_unrelated;
args.mode = C_UNRL;
init_hmm_unrelated(&args);
}
return 1;
}
int main_vcfquery(int argc, char *argv[])
{
int c, collapse = 0;
args_t *args = (args_t*) calloc(1,sizeof(args_t));
args->argc = argc; args->argv = argv;
int regions_is_file = 0, targets_is_file = 0;
static struct option loptions[] =
{
{"help",0,0,'h'},
{"list-samples",0,0,'l'},
{"include",1,0,'i'},
{"exclude",1,0,'e'},
{"format",1,0,'f'},
{"output-file",1,0,'o'},
{"regions",1,0,'r'},
{"regions-file",1,0,'R'},
{"targets",1,0,'t'},
{"targets-file",1,0,'T'},
{"annots",1,0,'a'},
{"samples",1,0,'s'},
{"samples-file",1,0,'S'},
{"print-header",0,0,'H'},
{"collapse",1,0,'c'},
{"vcf-list",1,0,'v'},
{"allow-undef-tags",0,0,'u'},
{0,0,0,0}
};
while ((c = getopt_long(argc, argv, "hlr:R:f:a:s:S:Ht:T:c:v:i:e:o:u",loptions,NULL)) >= 0) {
switch (c) {
case 'o': args->fn_out = optarg; break;
case 'f': args->format_str = strdup(optarg); break;
case 'H': args->print_header = 1; break;
case 'v': args->vcf_list = optarg; break;
case 'c':
if ( !strcmp(optarg,"snps") ) collapse |= COLLAPSE_SNPS;
else if ( !strcmp(optarg,"indels") ) collapse |= COLLAPSE_INDELS;
else if ( !strcmp(optarg,"both") ) collapse |= COLLAPSE_SNPS | COLLAPSE_INDELS;
else if ( !strcmp(optarg,"any") ) collapse |= COLLAPSE_ANY;
else if ( !strcmp(optarg,"all") ) collapse |= COLLAPSE_ANY;
else if ( !strcmp(optarg,"some") ) collapse |= COLLAPSE_SOME;
else error("The --collapse string \"%s\" not recognised.\n", optarg);
break;
case 'a':
{
kstring_t str = {0,0,0};
kputs("%CHROM\t%POS\t%MASK\t%REF\t%ALT\t%", &str);
char *p = optarg;
while ( *p )
{
if ( *p==',' )
kputs("\t%", &str);
else
kputc(*p, &str);
p++;
}
kputc('\n', &str);
args->format_str = str.s;
break;
}
case 'e': args->filter_str = optarg; args->filter_logic |= FLT_EXCLUDE; break;
case 'i': args->filter_str = optarg; args->filter_logic |= FLT_INCLUDE; break;
case 'r': args->regions_list = optarg; break;
case 'R': args->regions_list = optarg; regions_is_file = 1; break;
case 't': args->targets_list = optarg; break;
case 'T': args->targets_list = optarg; targets_is_file = 1; break;
case 'l': args->list_columns = 1; break;
case 'u': args->allow_undef_tags = 1; break;
case 's': args->sample_list = optarg; break;
case 'S': args->sample_list = optarg; args->sample_is_file = 1; break;
case 'h':
case '?': usage();
default: error("Unknown argument: %s\n", optarg);
}
}
char *fname = NULL;
if ( optind>=argc )
{
if ( !isatty(fileno((FILE *)stdin)) ) fname = "-";
}
else fname = argv[optind];
if ( args->list_columns )
{
if ( !fname ) error("Missing the VCF file name\n");
args->files = bcf_sr_init();
if ( !bcf_sr_add_reader(args->files, fname) ) error("Failed to open %s: %s\n", fname,bcf_sr_strerror(args->files->errnum));
list_columns(args);
bcf_sr_destroy(args->files);
free(args);
return 0;
}
if ( !args->format_str ) usage();
args->out = args->fn_out ? fopen(args->fn_out, "w") : stdout;
if ( !args->out ) error("%s: %s\n", args->fn_out,strerror(errno));
if ( !args->vcf_list )
{
if ( !fname ) usage();
args->files = bcf_sr_init();
args->files->collapse = collapse;
if ( optind+1 < argc ) args->files->require_index = 1;
if ( args->regions_list && bcf_sr_set_regions(args->files, args->regions_list, regions_is_file)<0 )
error("Failed to read the regions: %s\n", args->regions_list);
if ( args->targets_list )
{
if ( bcf_sr_set_targets(args->files, args->targets_list, targets_is_file, 0)<0 )
error("Failed to read the targets: %s\n", args->targets_list);
}
while ( fname )
{
if ( !bcf_sr_add_reader(args->files, fname) ) error("Failed to open %s: %s\n", fname,bcf_sr_strerror(args->files->errnum));
fname = ++optind < argc ? argv[optind] : NULL;
}
init_data(args);
query_vcf(args);
free(args->format_str);
destroy_data(args);
bcf_sr_destroy(args->files);
fclose(args->out);
free(args);
return 0;
}
// multiple VCFs
int i, k, nfiles, prev_nsamples = 0;
char **fnames, **prev_samples = NULL;
fnames = hts_readlist(args->vcf_list, 1, &nfiles);
if ( !nfiles ) error("No files in %s?\n", args->vcf_list);
for (i=0; i<nfiles; i++)
{
args->files = bcf_sr_init();
args->files->collapse = collapse;
if ( args->regions_list && bcf_sr_set_regions(args->files, args->regions_list, regions_is_file)<0 )
error("Failed to read the regions: %s\n", args->regions_list);
if ( optind < argc ) args->files->require_index = 1;
if ( args->targets_list )
{
if ( bcf_sr_set_targets(args->files, args->targets_list,targets_is_file, 0)<0 )
error("Failed to read the targets: %s\n", args->targets_list);
}
if ( !bcf_sr_add_reader(args->files, fnames[i]) ) error("Failed to open %s: %s\n", fnames[i],bcf_sr_strerror(args->files->errnum));
for (k=optind; k<argc; k++)
if ( !bcf_sr_add_reader(args->files, argv[k]) ) error("Failed to open %s: %s\n", argv[k],bcf_sr_strerror(args->files->errnum));
init_data(args);
if ( i==0 )
prev_samples = copy_header(args->header, args->files->readers[0].header->samples, bcf_hdr_nsamples(args->files->readers[0].header));
else
{
args->print_header = 0;
if ( compare_header(args->header, args->files->readers[0].header->samples, bcf_hdr_nsamples(args->files->readers[0].header), prev_samples, prev_nsamples) )
error("Different samples in %s and %s\n", fnames[i-1],fnames[i]);
}
query_vcf(args);
destroy_data(args);
bcf_sr_destroy(args->files);
}
fclose(args->out);
destroy_list(fnames, nfiles);
destroy_list(prev_samples, prev_nsamples);
free(args->format_str);
free(args);
return 0;
}
/*
* Reads sample names and their ploidy (optional) from a file.
* Alternatively, if no such file exists, the file name is interpreted
* as a comma-separated list of samples. When ploidy is not present,
* the default ploidy 2 is assumed.
*/
static void set_samples(args_t *args, const char *fn, int is_file)
{
int i, nlines;
char **lines = hts_readlist(fn, is_file, &nlines);
if ( !lines ) error("Could not read the file: %s\n", fn);
int nsmpls;
char **smpls = parse_ped_samples(&args->aux, lines, nlines, &nsmpls);
if ( smpls )
{
for (i=0; i<nlines; i++) free(lines[i]);
free(lines);
lines = smpls;
nlines = nsmpls;
}
args->samples_map = (int*) malloc(sizeof(int)*bcf_hdr_nsamples(args->aux.hdr)); // for subsetting
args->sample2sex = (int*) malloc(sizeof(int)*bcf_hdr_nsamples(args->aux.hdr));
int dflt_sex_id = ploidy_add_sex(args->ploidy, "F");
for (i=0; i<bcf_hdr_nsamples(args->aux.hdr); i++) args->sample2sex[i] = dflt_sex_id;
int *old2new = (int*) malloc(sizeof(int)*bcf_hdr_nsamples(args->aux.hdr));
for (i=0; i<bcf_hdr_nsamples(args->aux.hdr); i++) old2new[i] = -1;
int nsmpl = 0, map_needed = 0;
for (i=0; i<nlines; i++)
{
char *ss = lines[i];
while ( *ss && isspace(*ss) ) ss++;
if ( !*ss ) error("Could not parse: %s\n", lines[i]);
if ( *ss=='#' ) continue;
char *se = ss;
while ( *se && !isspace(*se) ) se++;
char x = *se, *xptr = se; *se = 0;
int ismpl = bcf_hdr_id2int(args->aux.hdr, BCF_DT_SAMPLE, ss);
if ( ismpl < 0 ) { fprintf(stderr,"Warning: No such sample in the VCF: %s\n",ss); continue; }
ss = se+1;
while ( *ss && isspace(*ss) ) ss++;
if ( !*ss ) ss = "2"; // default ploidy
se = ss;
while ( *se && !isspace(*se) ) se++;
if ( se==ss ) { *xptr = x; error("Could not parse: \"%s\"\n", lines[i]); }
if ( ss[1]==0 && (ss[0]=='0' || ss[0]=='1' || ss[0]=='2') )
args->sample2sex[nsmpl] = -1*(ss[0]-'0');
else
args->sample2sex[nsmpl] = ploidy_add_sex(args->ploidy, ss);
if ( ismpl!=nsmpl ) map_needed = 1;
args->samples_map[nsmpl] = ismpl;
old2new[ismpl] = nsmpl;
nsmpl++;
}
for (i=0; i<args->aux.nfams; i++)
{
int j, nmiss = 0;
family_t *fam = &args->aux.fams[i];
for (j=0; j<3; j++)
{
fam->sample[i] = old2new[fam->sample[i]];
if ( fam->sample[i]<0 ) nmiss++;
}
assert( nmiss==0 || nmiss==3 );
}
free(old2new);
if ( !map_needed ) { free(args->samples_map); args->samples_map = NULL; }
args->nsamples = nsmpl;
args->samples = lines;
}