static int set_sample_pairs(char **samples, int nsamples, kstring_t *hdr, int idx)
{
int i, j, n;
// Are these samples "old-name new-name" pairs?
void *hash = khash_str2str_init();
for (i=0; i<nsamples; i++)
{
char *key, *value;
key = value = samples[i];
while ( *value && !isspace(*value) ) value++;
if ( !*value ) break;
*value = 0; value++;
while ( isspace(*value) ) value++;
khash_str2str_set(hash,key,value);
}
if ( i!=nsamples ) // not "old-name new-name" pairs
{
khash_str2str_destroy(hash);
return 0;
}
while ( hdr->l>0 && isspace(hdr->s[hdr->l-1]) ) hdr->l--; // remove trailing newlines
hdr->s[hdr->l] = 0;
kstring_t tmp = {0,0,0};
i = j = n = 0;
while ( hdr->s[idx+i] && hdr->s[idx+i])
{
if ( hdr->s[idx+i]=='\t' )
{
hdr->s[idx+i] = 0;
if ( ++n>9 )
{
char *ori = khash_str2str_get(hash,hdr->s+idx+j);
kputs(ori ? ori : hdr->s+idx+j, &tmp);
}
else
kputs(hdr->s+idx+j, &tmp);
kputc('\t',&tmp);
j = ++i;
continue;
}
i++;
}
char *ori = khash_str2str_get(hash,hdr->s+idx+j);
kputs(ori ? ori : hdr->s+idx+j, &tmp);
if ( hash ) khash_str2str_destroy(hash);
hdr->l = idx;
kputs(tmp.s, hdr);
kputc('\n', hdr);
free(tmp.s);
return 1;
}
static int bsmpl_keep_readgroup(bam_smpl_t *bsmpl, file_t *file, const char *rg_id, const char **smpl_name)
{
char *rg_smpl = khash_str2str_get(bsmpl->rg_list,rg_id); // unique read group present in one bam only
if ( !rg_smpl )
{
// read group specific to this bam
bsmpl->tmp.l = 0;
ksprintf(&bsmpl->tmp,"%s\t%s",rg_id,file->fname);
rg_smpl = khash_str2str_get(bsmpl->rg_list,bsmpl->tmp.s);
}
if ( !rg_smpl )
{
// any read group in this file?
bsmpl->tmp.l = 0;
ksprintf(&bsmpl->tmp,"*\t%s",file->fname);
rg_smpl = khash_str2str_get(bsmpl->rg_list,bsmpl->tmp.s);
}
if ( !rg_smpl && bsmpl->rg_logic ) return 0;
if ( rg_smpl && !bsmpl->rg_logic ) return 0;
if ( rg_smpl && rg_smpl[0]!='\t' ) *smpl_name = rg_smpl; // rename the sample
return 1;
}
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;
}
/*
The logic of this function is a bit complicated because we want to work
also with broken bams containing read groups that are not listed in the
header. The desired behavior is as follows:
- when -G is given, read groups which are not listed in the header must
be given explicitly using the "?" symbol in -G.
Otherwise:
- if the bam has no header, all reads in the file are assigned to a
single sample named after the file
- if there is at least one sample defined in the header, reads with no
read group id or with a read group id not listed in the header are
assigned to the first sample encountered in the header
*/
int bam_smpl_add_bam(bam_smpl_t *bsmpl, char *bam_hdr, const char *fname)
{
bsmpl->nfiles++;
bsmpl->files = (file_t*) realloc(bsmpl->files,bsmpl->nfiles*sizeof(file_t));
file_t *file = &bsmpl->files[bsmpl->nfiles-1];
memset(file,0,sizeof(file_t));
file->fname = strdup(fname);
file->default_idx = -1;
if ( bsmpl->ignore_rg || !bam_hdr )
{
// The option --ignore-RG is set or there is no BAM header: use the file name as the sample name
bsmpl_add_readgroup(bsmpl,file,"*",file->fname);
return bsmpl->nfiles-1;
}
void *bam_smpls = khash_str2int_init();
int first_smpl = -1, nskipped = 0;
const char *p = bam_hdr, *q, *r;
while (p != NULL && (q = strstr(p, "@RG")) != 0)
{
char *eol = strchr(q + 3, '\n');
if (q > bam_hdr && *(q - 1) != '\n') { // @RG must be at start of line
p = eol;
continue;
}
p = q + 3;
if ((q = strstr(p, "\tID:")) != 0) q += 4;
if ((r = strstr(p, "\tSM:")) != 0) r += 4;
if (r && q)
{
char *u, *v;
int ioq, ior;
for (u = (char*)q; *u && *u != '\t' && *u != '\n'; ++u);
for (v = (char*)r; *v && *v != '\t' && *v != '\n'; ++v);
ioq = *u; ior = *v; *u = *v = '\0';
// q now points to a null terminated read group id
// r points to a null terminated sample name
if ( !strcmp("*",q) || !strcmp("?",q) )
error("Error: the read group IDs \"*\" and \"?\" have a special meaning in the mpileup code. Please fix the code or the bam: %s\n", fname);
int accept_rg = 1;
if ( bsmpl->sample_list )
{
// restrict samples based on the -s/-S options
char *name = khash_str2str_get(bsmpl->sample_list,r);
if ( bsmpl->sample_logic==0 )
accept_rg = name ? 0 : 1;
else if ( !name )
accept_rg = 0;
else
r = name;
}
if ( accept_rg && bsmpl->rg_list )
{
// restrict readgroups based on the -G option, possibly renaming the sample
accept_rg = bsmpl_keep_readgroup(bsmpl,file,q,&r);
}
if ( accept_rg )
bsmpl_add_readgroup(bsmpl,file,q,r);
else
{
bsmpl_add_readgroup(bsmpl,file,q,NULL); // ignore this RG but note that it was seen in the header
nskipped++;
}
if ( first_smpl<0 )
khash_str2int_get(bsmpl->name2idx,r,&first_smpl);
if ( !khash_str2int_has_key(bam_smpls,r) )
khash_str2int_inc(bam_smpls,strdup(r));
*u = ioq; *v = ior;
}
else
break;
p = eol;
}
int nsmpls = khash_str2int_size(bam_smpls);
khash_str2int_destroy_free(bam_smpls);
const char *smpl_name = NULL;
int accept_null_rg = 1;
if ( bsmpl->rg_list && !bsmpl_keep_readgroup(bsmpl,file,"?",&smpl_name) ) accept_null_rg = 0;
if ( bsmpl->sample_list && first_smpl==-1 ) accept_null_rg = 0;
if ( !accept_null_rg && first_smpl==-1 )
{
// no suitable read group is available in this bam: ignore the whole file.
free(file->fname);
if ( file->rg2idx ) khash_str2int_destroy_free(file->rg2idx);
bsmpl->nfiles--;
return -1;
}
if ( !accept_null_rg ) return bsmpl->nfiles-1;
if ( nsmpls==1 && !nskipped )
{
file->default_idx = first_smpl;
return bsmpl->nfiles-1;
}
if ( !smpl_name ) smpl_name = first_smpl==-1 ? file->fname : bsmpl->smpl[first_smpl];
bsmpl_add_readgroup(bsmpl,file,"?",smpl_name);
return bsmpl->nfiles-1;
}