// FIXME: we should also check the LB tag associated with each alignment
const char *bam_get_library(bam_header_t *h, const bam1_t *b)
{
const uint8_t *rg;
if (h->dict == 0) h->dict = sam_header_parse2(h->text);
if (h->rg2lib == 0) h->rg2lib = sam_header2tbl(h->dict, "RG", "ID", "LB");
rg = bam_aux_get(b, "RG");
return (rg == 0)? 0 : sam_tbl_get(h->rg2lib, (const char*)(rg + 1));
}
bam_header_t *bam_header_dup(const bam_header_t *h0)
{
bam_header_t *h;
int i;
h = bam_header_init();
*h = *h0;
h->hash = h->rg2lib = 0;
h->header = 0;
h->text = (char*)calloc(h->l_text + 1, 1);
memcpy(h->text, h0->text, h->l_text);
h->target_len = (uint32_t*)calloc(h->n_targets, 4);
h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
for (i = 0; i < h->n_targets; ++i) {
h->target_len[i] = h0->target_len[i];
h->target_name[i] = strdup(h0->target_name[i]);
}
if (h->header == 0) h->header = sam_header_parse2(h->text);
return h;
}
samfile_t* b2g_samfile_open(char* path, char* mode, void* aux, ...) {
char filename[1024] = {};
va_list args;
va_start(args, aux);
vsprintf(filename, path, args);
va_end(args);
if ('r' == mode[0] && access(filename, R_OK)) return NULL;
samfile_t *in = samopen(filename, mode, aux);
if (in) {
if (!in->header) return NULL;
// Make sure to parse the header for later use.
bam_header_t *header = in->header;
if (header->dict == 0) header->dict = sam_header_parse2(header->text);
if (header->rg2lib == 0) header->rg2lib = sam_header2tbl(header->dict, "RG", "ID", "LB");
}
return in;
}
bam_header_t *
tmap_seqs_io_to_bam_header(tmap_refseq_t *refseq,
tmap_seqs_io_t *io_in,
char **rg_sam, int32_t rg_sam_num,
int32_t argc, char *argv[])
{
bam_header_t *bam_header = NULL;
sam_header_t *header = NULL; // the output header
sam_header_record_t *record = NULL;
sam_header_record_t **record_list = NULL;
char tag[2];
char *command_line= NULL;
char *id = NULL;
char *id_pp = NULL;
int32_t i, j;
// @HD
if(io_in->type == TMAP_SEQ_TYPE_SAM || io_in->type == TMAP_SEQ_TYPE_BAM) {
// should be only one input file
if(1 != io_in->n) {
tmap_bug();
}
// get the current header
if(NULL == io_in->seqios[0]) tmap_bug();
if(NULL == io_in->seqios[0]->io.samio) tmap_bug();
if(NULL == io_in->seqios[0]->io.samio->fp->header) tmap_bug();
if(NULL == io_in->seqios[0]->io.samio->fp->header->header) {
header = sam_header_parse2(io_in->seqios[0]->io.samio->fp->header->text);
}
else {
header = io_in->seqios[0]->io.samio->fp->header->header; // wow, that's a lot of pointers
if(NULL == header) tmap_bug();
header = sam_header_clone(header); // clone the header
}
if(NULL == header) tmap_bug();
}
else {
// empty header
header = sam_header_init();
// @HD - header line
record = sam_header_record_init("HD"); // new header line
if(0 == sam_header_record_add(record, "VN", "1.4")) tmap_bug(); // version number
if(0 == sam_header_add_record(header, record)) tmap_bug(); // add the header line
// nullify
record = NULL;
}
// Get the TMAP program ID
id = tmap_malloc(sizeof(char) * (1 + strlen(PACKAGE_NAME)), "id");
strcpy(id, PACKAGE_NAME); // default
for(i=j=0;NULL != (record_list = sam_header_get_record(header, "PG", "ID", id, &i)) && 0 < i;i=0) { // while the id is found
char *ptr = NULL;
// swap id and id_pp
ptr = id_pp;
id_pp = id;
id = ptr;
// create the new ID
j++;
id = tmap_realloc(id, sizeof(char) * (1 + (int)log10(j) + 1 + strlen(PACKAGE_NAME) + 1), "id");
if(sprintf(id, "%s.%d", PACKAGE_NAME, j) < 0) tmap_bug();
free(record_list);
record_list = NULL;
}
// @SQ
if(NULL != refseq) {
sam_header_records_t *records = NULL;
// NB: check to see if any SQ/SN records exist, if not, then ignore checking...
// ZZ: We will not checking, but instead just remove all the old header. The old way of checking is not working
records = sam_header_get_records(header, "SQ");
if (NULL != records) {
// ZZ: remove the headers if exists.
sam_header_remove_records(header, "SQ");
records = NULL;
}
// ZZ: Now we will just add all new tags
for(i=0;i<refseq->num_annos;i++) { // for each reference sequence
char num[32];
record = sam_header_record_init("SQ"); // new reference sequence record
if(0 == sam_header_record_add(record, "SN", refseq->annos[i].name->s)) tmap_bug(); // reference sequence name
if(sprintf(num, "%u", (uint32_t)refseq->annos[i].len) < 0) tmap_bug(); // integer to string
if(0 == sam_header_record_add(record, "LN", num)) tmap_bug(); // reference sequence length
if(0 == sam_header_add_record(header, record)) tmap_bug(); // add the reference sequence record
}
}
// @RG - read group
if(0 < rg_sam_num) { // @RG specified on the command line
// Check for SAM/BAM
// TODO: this should be possible...
if(io_in->type == TMAP_SEQ_TYPE_SAM || io_in->type == TMAP_SEQ_TYPE_BAM) {
tmap_error("Cannot specify the read groups on the command line when using SAM/BAM as input."
" Please embed in the SAM/BAM header instead.", Exit, OutOfRange);
}
record = NULL;
// go through the command line arguments
for(i=0;i<rg_sam_num;i++) {
if(strlen(rg_sam[i]) < 4) tmap_error("Read group too small", Exit, OutOfRange);
if(':' != rg_sam[i][2]) tmap_error("Read group improperly formatted (no colon)", Exit, OutOfRange);
// check for id
if('I' == rg_sam[i][0] && 'D' == rg_sam[i][1]) { // new read group
if(NULL != record) { // add the record
tmap_seqs_io_init2_fs_and_add(io_in, header, record); // add @RG.KS and @RG.FO
}
record = sam_header_record_init("RG"); // new read group
}
// add the tag/value to the record
if(NULL == record) {
tmap_error("The read group ID must be specified first", Exit, OutOfRange);
}
tag[0]=rg_sam[i][0]; tag[1]=rg_sam[i][1]; // setup the tag
if(0 == sam_header_record_add(record, tag, rg_sam[i]+3)) tmap_bug(); // add the tag/value
}
if(NULL != record) { // add the record
tmap_seqs_io_init2_fs_and_add(io_in, header, record); // add @RG.KS and @RG.FO
}
// check that the # of read groups added was the same as the # of input files...
sam_header_records_t *records = sam_header_get_records(header, "RG"); // get the header line
if(records->n != io_in->n) tmap_error("The number of read groups did not match the number of input files", Exit, OutOfRange);
}
else if(io_in->type != TMAP_SEQ_TYPE_SAM && io_in->type != TMAP_SEQ_TYPE_BAM) { // dummy...
for(i=0;i<io_in->n;i++) { // for each input file
char buf[32];
record = sam_header_record_init("RG"); // new read group
if(1 == io_in->n) strcpy(buf, "NOID");
else if(sprintf(buf, "NOID.%d", i+1) < 0) tmap_bug();
if(0 == sam_header_record_add(record, "ID", buf)) tmap_bug(); // dummy ID
if(0 == sam_header_record_add(record, "SM", "NOSM")) tmap_bug(); // dummy SM, for Picard validation
if(0 == sam_header_record_add(record, "PG", id)) tmap_bug(); // dummy PG
tmap_seqs_io_init2_fs_and_add(io_in, header, record); // add @RG.KS and @RG.FO
}
}
else {
// check that SM/PG are present
sam_header_records_t *records = sam_header_get_records(header, "RG"); // get the header line
for(i=0;i<records->n;i++) {
record = records->records[i];
if(NULL == sam_header_record_get(record, "ID")) tmap_error("Missing @RG.ID in the SAM/BAM Header", Exit, OutOfRange);
if(NULL == sam_header_record_get(record, "SM")) {
if(0 == sam_header_record_add(record, "SM", "NOSM")) tmap_bug(); // dummy SM, for Picard validation
}
if(NULL == sam_header_record_get(record, "PG")) {
if(0 == sam_header_record_add(record, "PG", id)) tmap_bug(); // dummy PG
}
}
}
// @PG - program group
// TODO: check for previous program group ID and set @PG.PP
record = sam_header_record_init("PG"); // new program group
if(0 == sam_header_record_add(record, "ID", id)) tmap_bug(); // @PG.ID
if(0 == sam_header_record_add(record, "VN", PACKAGE_VERSION)) tmap_bug(); // @PG.VN
// @PG.CL
command_line = NULL;
j = 1; // for the EOL
command_line = tmap_realloc(command_line, sizeof(char) * j, "command_line");
command_line[j-1] = '\0';
for(i=0;i<argc;i++) {
if(0 < i) j++;
j += strlen(argv[i]);
command_line = tmap_realloc(command_line, sizeof(char) * j, "command_line");
if(0 < i) strcat(command_line, " ");
strcat(command_line, argv[i]);
command_line[j-1] = '\0';
}
if(0 == sam_header_record_add(record, "CL", command_line)) tmap_bug(); // @PG.CL
if(NULL != id_pp) { // @PG.PP
if(0 == sam_header_record_add(record, "PP", id_pp)) tmap_bug(); // @PG.CL
}
if(0 == sam_header_add_record(header, record)) tmap_bug(); // add the record
free(command_line);
// Check the new SAM Header
if(0 == sam_header_check(header)) {
tmap_error("SAM Header was not consistent", Exit, OutOfRange);
}
// Create a BAM Header from the SAM Header
bam_header = bam_header_init(); // empty
bam_header->header = header; // soft-copy the header
bam_header = sam_header_to_bam_header(bam_header); // convert
// free memory
free(id);
free(id_pp);
return bam_header;
}
