READ_ALN* new_read_aln(hashtable ht,char*read_name) {
// look for read in HT
// add to hash table
READ_ALN* r;
if((r=get_read_aln(ht,read_name))==NULL) {
// Memory chunck: |[index_entry]len bytes+1|
int len=strlen(read_name);
ulong key=hashit(read_name);
char *mem_block=(char*)malloc(sizeof(READ_ALN)+len+1);
index_mem+=sizeof(READ_ALN)+len+1+sizeof(hashnode);
if (mem_block==NULL) {
fprintf(stderr,"Error allocating memory\n");
exit(1);
}
r=(READ_ALN*)&mem_block[0];
r->name=(char*)&mem_block[sizeof(READ_ALN)];
r->name[len]='\0';
strncpy(r->name,read_name,len+1);
r->ctr=0;
if(insere(ht,key,r)<0){
fprintf(stderr,"Error adding %s to index\n",read_name);
exit(1);
}
}
r->ctr++;
return(r);
}
int main(int argc, char *argv[])
{
short out2stdout=0;
hashtable ht=new_hashtable(HASHSIZE);
bamFile in,in2;
bamFile out;
int paired;//1 if not paired or pair read 1, 2 otherwise
index_mem=sizeof(hashtable)*sizeof(hashnode**)*HASHSIZE*2;
if (argc != 3) {
fprintf(stderr, "Usage: bam_fix_NH <in.bam> <out.bam or - for stdout>\n");
return 1;
}
// Open file and exit if error
in = bam_open(argv[1], "rb");
out2stdout = strcmp(argv[2], "-")? 0 : 1;
out = strcmp(argv[2], "-")? bam_open(argv[2], "w") : bam_dopen(fileno(stdout), "w");
if (in == 0 ) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[1]);
return 1;
}
if (out == 0) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[2]);
return 1;
}
unsigned long num_alns=0;
int ref;
// ***********
// Copy header
bam_header_t *header;
header = bam_header_read(in);
bam_header_write(out,header);
// sorted by name?
// Should not rely on the value in SO
bam1_t *aln=bam_init1();
bam1_t *prev=bam_init1();
if (!out2stdout) {
fprintf(stderr,"bam_fix_NH version %s\n",VERSION);
fprintf(stderr,"Processing %s\n",argv[1]);
fprintf(stderr,"Hashing...\n");fflush(stderr);
}
while(bam_read1(in,aln)>=0) { // read alignment
if (aln->core.tid < 0) continue;//ignore unaligned reads
if (aln->core.flag & BAM_FUNMAP) continue;
if (aln->core.flag & BAM_FREAD2) paired=2;
else paired=1;
++num_alns;
new_read_aln(ht,fix_read_name(bam1_qname(aln),paired));
if(!out2stdout) PRINT_ALNS_PROCESSED(num_alns);
}
bam_close(in);
if(!out2stdout) {
fprintf(stderr,"%s%lu\n",BACKLINE,num_alns);
fprintf(stderr,"Hashing complete (%lu alignments)\n",num_alns);
fprintf(stderr,"Memory used: %ld MB\n",index_mem/1024/1024);
fprintf(stderr,"Updating entries with NH and printing BAM...\n");
fflush(stderr);
}
// reopen
in2 = bam_open(argv[1], "rb");
if (in2 == 0 ) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[1]);
return 1;
}
header = bam_header_read(in2);
num_alns=0;
while(bam_read1(in2,aln)>=0) { // read alignment
paired=1;
if (aln->core.tid < 0) continue;//ignore unaligned reads
if (aln->core.flag & BAM_FUNMAP) continue;
if (aln->core.flag & BAM_FREAD2) paired=2;
++num_alns;
READ_ALN *r=get_read_aln(ht,fix_read_name(bam1_qname(aln),paired));
assert(r!=NULL);
// update the NH field
uint8_t *old_nh = bam_aux_get(aln, "NH");
int32_t nh=r->ctr;
if (old_nh) {
if (nh!=bam_aux2i(old_nh)) {
fprintf(stderr,"warning: value mismatch! replacing>%s %d->%d\n",bam1_qname(aln),bam_aux2i(old_nh),nh);
}
bam_aux_del(aln, old_nh);
bam_aux_append(aln, "NH", 'i', 4, (uint8_t*)&nh);
#ifdef DEBUG
// printf("!>%s %d\n",bam1_qname(aln),r->ctr);
#endif
}
if (!old_nh) { // add NH
bam_aux_append(aln, "NH", 'i', 4, (uint8_t*)&nh);
#ifdef DEBUG
fprintf(stderr,"!>%s %d\n",bam1_qname(aln),bam_aux2i(old_nh));
#endif
}
bam_write1(out,aln);
if(!out2stdout) PRINT_ALNS_PROCESSED(num_alns);
}
//
bam_destroy1(aln);
bam_close(in2);
bam_close(out);
if(!out2stdout) {
fprintf(stderr,"%s%lu\n",BACKLINE,num_alns);
fprintf(stderr,"Done.\n");
}
return 0;
}
int main(int argc, char *argv[])
{
hashtable ht=new_hashtable(HASHSIZE);
bamFile in,in2;
bamFile out;
if (argc != 3) {
fprintf(stderr, "Usage: bam_fix_NH <in.bam> <out.bam>\n");
return 1;
}
// Open file and exit if error
//in = strcmp(argv[1], "-")? bam_open(argv[1], "rb") : bam_dopen(fileno(stdin), "rb");
in = bam_open(argv[1], "rb");
out = strcmp(argv[2], "-")? bam_open(argv[2], "w") : bam_dopen(fileno(stdout), "w");
if (in == 0 ) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[1]);
return 1;
}
if (out == 0) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[2]);
return 1;
}
unsigned long num_alns=0;
int ref;
// ***********
// Copy header
bam_header_t *header;
header = bam_header_read(in);
bam_header_write(out,header);
// sorted by name?
// Should not rely on the value in SO
bam1_t *aln=bam_init1();
bam1_t *prev=bam_init1();
printf("Hashing...\n");flush(stdout);
while(bam_read1(in,aln)>=0) { // read alignment
if (aln->core.tid < 0) continue;//ignore unaligned reads
++num_alns;
new_read_aln(ht,bam1_qname(aln));
}
bam_close(in);
printf("Hashing complete (%lu alignments)\n",num_alns);
printf("Memory used in the hash: %ld MB\n",index_mem/1024/1024);
flush(stdout);
// reopen
in2 = bam_open(argv[1], "rb");
if (in2 == 0 ) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[1]);
return 1;
}
header = bam_header_read(in2);
while(bam_read1(in2,aln)>=0) { // read alignment
if (aln->core.tid < 0) continue;//ignore unaligned reads
++num_alns;
READ_ALN *r=get_read_aln(ht,bam1_qname(aln));
//assert(r!=NULL);
// update the NH field
uint8_t *old_nh = bam_aux_get(aln, "NH");
uint8_t nh=r->ctr;
if (old_nh) {
if (nh!=bam_aux2i(old_nh)) {
fprintf(stderr,"warning: value mismatch! replacing>%s %d->%d\n",bam1_qname(aln),bam_aux2i(old_nh),nh);
}
bam_aux_del(aln, old_nh);
bam_aux_append(aln, "NH", 'i', 4, (uint8_t*)&nh);
}
if (!old_nh) { // add NH
bam_aux_append(aln, "NH", 'i', 4, (uint8_t*)&nh);
#ifdef DEBUG
printf("!>%s %d\n",bam1_qname(aln),bam_aux2i(old_nh));
#endif
}
// in->header
// Also fix the XS:A tag
// BAM_FREAD1
// BAM_FREAD2
// BAM_FREVERSE the read is mapped to the reverse strand
//bam1_cigar(b)
//BAM_CREF_SKIP 3 CIGAR skip on the reference (e.g. spliced alignment)
//BAM_FREVERSE 16 the read is mapped to the reverse strand
if (aln->core.flag & BAM_FSECONDARY) continue; // skip secondary alignments
if (aln->core.flag & ! BAM_FPAIRED) continue; // not paired
if (aln->core.flag & ! BAM_FPROPER_PAIR) continue; // not a proper pair
if (aln->core.flag & ! BAM_FMUNMAP) continue; // the mate is mapped
if (aln->core.flag & BAM_FSECONDARY) continue; // secundary read
if (aln->core.flag & BAM_FREAD2) continue; // only count each pair once
// core.strand == 0 (f/+) 1 r/-
// flag
// bam1_qname(b)
bam_write1(out,aln);
}
//
bam_destroy1(aln);
bam_close(in2);
bam_close(out);
return 0;
/*
uint8_t *old_nm = bam_aux_get(b, "NM");
90 if (c->flag & BAM_FUNMAP) return;
91 if (old_nm) old_nm_i = bam_aux2i(old_nm);
92 if (!old_nm) bam_aux_append(b, "NM", 'i', 4, (uint8_t*)&nm);
93 else if (nm != old_nm_i) {
94 fprintf(stderr, "[bam_fillmd1] different NM for read '%s': %d -> %d\n", bam1_qname(b), old_nm_i, nm);
95 bam_aux_del(b, old_nm);
96 bam_aux_append(b, "NM", 'i', 4, (uint8_t*)&nm);
97 }
*/
}