int faidx_main(int argc, char *argv[])
{
if (argc == 1) {
fprintf(stderr, "Usage: faidx <in.fasta> [<reg> [...]]\n");
return 1;
} else {
if (argc == 2) fai_build(argv[1]);
else {
int i, j, k, l;
char *s;
faidx_t *fai;
fai = fai_load(argv[1]);
if (fai == 0) return 1;
for (i = 2; i != argc; ++i) {
printf(">%s\n", argv[i]);
s = fai_fetch(fai, argv[i], &l);
for (j = 0; j < l; j += 60) {
for (k = 0; k < 60 && k < l - j; ++k)
putchar(s[j + k]);
putchar('\n');
}
free(s);
}
fai_destroy(fai);
}
}
return 0;
}
int get_unpadded_len(faidx_t *fai, char *ref_name, int padded_len)
{
char base;
char *fai_ref = 0;
int fai_ref_len = 0, k;
int bases=0, gaps=0;
fai_ref = fai_fetch(fai, ref_name, &fai_ref_len);
if (fai_ref_len != padded_len) {
fprintf(stderr, "[depad] ERROR: FASTA sequence '%s' length %i, expected %i\n", ref_name, fai_ref_len, padded_len);
free(fai_ref);
return -1;
}
for (k = 0; k < padded_len; ++k) {
//fprintf(stderr, "[depad] checking base %i of %i or %i\n", k+1, ref_len, strlen(fai_ref));
base = fai_ref[k];
if (base == '-' || base == '*') {
gaps += 1;
} else {
int i = seq_nt16_table[(int)base];
if (i == 0 || i==16) { // Equals maps to 0, anything unexpected to 16
fprintf(stderr, "[depad] ERROR: Invalid character %c (ASCII %i) in FASTA sequence '%s'\n", base, (int)base, ref_name);
free(fai_ref);
return -1;
}
bases += 1;
}
}
free(fai_ref);
assert (padded_len == bases + gaps);
return bases;
}
int load_unpadded_ref(faidx_t *fai, char *ref_name, int ref_len, kstring_t *seq)
{
char base;
char *fai_ref = 0;
int fai_ref_len = 0, k;
fai_ref = fai_fetch(fai, ref_name, &fai_ref_len);
if (fai_ref_len != ref_len) {
fprintf(stderr, "[depad] ERROR: FASTA sequence %s length %i, expected %i\n", ref_name, fai_ref_len, ref_len);
free(fai_ref);
return -1;
}
ks_resize(seq, ref_len);
seq->l = 0;
for (k = 0; k < ref_len; ++k) {
base = fai_ref[k];
if (base == '-' || base == '*') {
// Map gaps to null to match unpad_seq function
seq->s[seq->l++] = 0;
} else {
int i = seq_nt16_table[(int)base];
if (i == 0 || i==16) { // Equals maps to 0, anything unexpected to 16
fprintf(stderr, "[depad] ERROR: Invalid character %c (ASCII %i) in FASTA sequence %s\n", base, (int)base, ref_name);
free(fai_ref);
return -1;
}
seq->s[seq->l++] = i;
}
}
assert(ref_len == seq->l);
free(fai_ref);
return 0;
}
std::string IndexedFastaReader::Subsequence(const char *htslibRegion) const
{
REQUIRE_FAIDX_LOADED;
int len;
char* rawSeq = fai_fetch(handle_, htslibRegion, &len);
if (rawSeq == nullptr)
throw std::runtime_error("could not fetch FASTA sequence");
else {
std::string seq(rawSeq);
free(rawSeq);
return seq;
}
}
// get referense sequence
// NOTE: 1-based positions
std::string BasicHaplotypeBuilder::getRefSeq(std::string chr, int lpos, int rpos) {
if (!fai) {
throw std::string("FAI error.");
}
std::stringstream ss;
ss << chr << ":" << lpos << "-" << rpos;
char *ref;
int len;
ref = fai_fetch(fai, ss.str().c_str(), &len);
if (len == 0) {
throw std::string("faidx error: len==0");
}
std::string res(ref);
free(ref);
std::transform(res.begin(), res.end(), res.begin(), ::toupper);
return res;
}
std::string ReadDB::get_read_sequence(const std::string& read_id) const
{
assert(m_fai != NULL);
int length;
char* seq;
// this call is not threadsafe
#pragma omp critical
seq = fai_fetch(m_fai, read_id.c_str(), &length);
if(seq == NULL) {
return "";
}
std::string out(seq);
free(seq);
return out;
}
int bam_fillmd(int argc, char *argv[])
{
int c, flt_flag, tid = -2, ret, len, is_bam_out, is_uncompressed, max_nm, is_realn, capQ, baq_flag;
samFile *fp = NULL, *fpout = NULL;
bam_hdr_t *header = NULL;
faidx_t *fai = NULL;
char *ref = NULL, mode_w[8], *ref_file;
bam1_t *b = NULL;
sam_global_args ga = SAM_GLOBAL_ARGS_INIT;
static const struct option lopts[] = {
SAM_OPT_GLOBAL_OPTIONS('-', 0, 0, 0, 0),
{ NULL, 0, NULL, 0 }
};
flt_flag = UPDATE_NM | UPDATE_MD;
is_bam_out = is_uncompressed = is_realn = max_nm = capQ = baq_flag = 0;
strcpy(mode_w, "w");
while ((c = getopt_long(argc, argv, "EqreuNhbSC:n:Ad", lopts, NULL)) >= 0) {
switch (c) {
case 'r':
is_realn = 1;
break;
case 'e':
flt_flag |= USE_EQUAL;
break;
case 'd':
flt_flag |= DROP_TAG;
break;
case 'q':
flt_flag |= BIN_QUAL;
break;
case 'h':
flt_flag |= HASH_QNM;
break;
case 'N':
flt_flag &= ~(UPDATE_MD|UPDATE_NM);
break;
case 'b':
is_bam_out = 1;
break;
case 'u':
is_uncompressed = is_bam_out = 1;
break;
case 'S':
break;
case 'n':
max_nm = atoi(optarg);
break;
case 'C':
capQ = atoi(optarg);
break;
case 'A':
baq_flag |= 1;
break;
case 'E':
baq_flag |= 2;
break;
default:
if (parse_sam_global_opt(c, optarg, lopts, &ga) == 0) break;
fprintf(stderr, "[bam_fillmd] unrecognized option '-%c'\n\n", c);
/* else fall-through */
case '?':
return calmd_usage();
}
}
if (is_bam_out) strcat(mode_w, "b");
else strcat(mode_w, "h");
if (is_uncompressed) strcat(mode_w, "0");
if (optind + (ga.reference == NULL) >= argc)
return calmd_usage();
fp = sam_open_format(argv[optind], "r", &ga.in);
if (fp == NULL) {
print_error_errno("calmd", "Failed to open input file '%s'", argv[optind]);
return 1;
}
header = sam_hdr_read(fp);
if (header == NULL || header->n_targets == 0) {
fprintf(stderr, "[bam_fillmd] input SAM does not have header. Abort!\n");
goto fail;
}
fpout = sam_open_format("-", mode_w, &ga.out);
if (fpout == NULL) {
print_error_errno("calmd", "Failed to open output");
goto fail;
}
if (sam_hdr_write(fpout, header) < 0) {
print_error_errno("calmd", "Failed to write sam header");
goto fail;
}
ref_file = argc > optind + 1 ? argv[optind+1] : ga.reference;
fai = fai_load(ref_file);
if (!fai) {
print_error_errno("calmd", "Failed to open reference file '%s'", ref_file);
goto fail;
}
b = bam_init1();
if (!b) {
fprintf(stderr, "[bam_fillmd] Failed to allocate bam struct\n");
goto fail;
}
while ((ret = sam_read1(fp, header, b)) >= 0) {
if (b->core.tid >= 0) {
if (tid != b->core.tid) {
free(ref);
ref = fai_fetch(fai, header->target_name[b->core.tid], &len);
tid = b->core.tid;
if (ref == 0) { // FIXME: Should this always be fatal?
fprintf(stderr, "[bam_fillmd] fail to find sequence '%s' in the reference.\n",
header->target_name[tid]);
if (is_realn || capQ > 10) goto fail; // Would otherwise crash
}
}
if (is_realn) sam_prob_realn(b, ref, len, baq_flag);
if (capQ > 10) {
int q = sam_cap_mapq(b, ref, len, capQ);
if (b->core.qual > q) b->core.qual = q;
}
if (ref) bam_fillmd1_core(b, ref, len, flt_flag, max_nm);
}
if (sam_write1(fpout, header, b) < 0) {
print_error_errno("calmd", "failed to write to output file");
goto fail;
}
}
if (ret < -1) {
fprintf(stderr, "[bam_fillmd] Error reading input.\n");
goto fail;
}
bam_destroy1(b);
bam_hdr_destroy(header);
free(ref);
fai_destroy(fai);
sam_close(fp);
if (sam_close(fpout) < 0) {
fprintf(stderr, "[bam_fillmd] error when closing output file\n");
return 1;
}
return 0;
fail:
free(ref);
if (b) bam_destroy1(b);
if (header) bam_hdr_destroy(header);
if (fai) fai_destroy(fai);
if (fp) sam_close(fp);
if (fpout) sam_close(fpout);
return 1;
}
static int dindel_fetch_func(bam1_t *b, void *data)
{
data_t_dindel *tmp = (data_t_dindel*)data;
bam1_core_t *c = &b->core;
int rlen;
uint8_t *to_delete;
/* don't change reads failing default mask: BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP */
if (c->flag & BAM_DEF_MASK) {
/* fprintf(stderr, "skipping read: %s at pos %d\n", bam1_qname(b), c->pos); */
bam_write1(tmp->out, b);
return 0;
}
/* get the reference sequence and compute homopolymer array */
if (tmp->tid != c->tid) {
/*fprintf(stderr, "fetching reference sequence %s\n",
tmp->in->header->target_name[c->tid]); */
char *ref = fai_fetch(tmp->fai, tmp->in->header->target_name[c->tid], &rlen);
strtoupper(ref);/* safeguard */
int rlen = strlen(ref);
tmp->tid = c->tid;
if (tmp->hpcount) free(tmp->hpcount);
tmp->hpcount = (int*)malloc(rlen*sizeof(int));
find_homopolymers(ref, tmp->hpcount, rlen);
free(ref);
tmp->rlen = rlen;
/* fprintf(stderr, "fetched reference sequence\n");*/
}
/* parse the cigar string */
uint32_t *cigar = bam1_cigar(b);
uint8_t indelq[c->l_qseq+1];
/* fprintf(stderr, "l_qseq:%d\n", c->l_qseq); */
int i;
int x = c->pos; /* coordinate on reference */
int y = 0; /* coordinate on query */
for (i = 0; i < c->n_cigar; ++i) {
int j, oplen = cigar[i]>>4, op = cigar[i]&0xf;
if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
for (j = 0; j < oplen; j++) {
/*fprintf(stderr, "query:%d, ref:%d, count:%d\n",
y, x, tmp->hpcount[x+1]); */
/* FIXME clang complains: The left operand of '>' is a garbage value */
indelq[y] = (x > tmp->rlen-2) ? DINDELQ[0] : (tmp->hpcount[x+1]>18 ?
DINDELQ[0] : DINDELQ[tmp->hpcount[x+1]]);
x++;
y++;
}
} else if (op == BAM_CHARD_CLIP) { /* do nothing */
} else if (op == BAM_CDEL) {
x += oplen;
} else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) {
for (j = 0; j < oplen; j++) {
/* fprintf(stderr, "query:%d, ref:%d\n", y, x); */
indelq[y] = DINDELQ[0];
y++;
}
} else {
LOG_FATAL("unknown op %d for read %s\n", op, bam1_qname(b));/* FIXME skip? seen this somewhere else properly handled */
exit(1);
}
}
indelq[y] = '\0';
to_delete = bam_aux_get(b, BI_TAG);
if (to_delete) {
bam_aux_del(b, to_delete);
}
bam_aux_append(b, BI_TAG, 'Z', c->l_qseq+1, indelq);
to_delete = bam_aux_get(b, BD_TAG);
if (to_delete) {
bam_aux_del(b, to_delete);
}
bam_aux_append(b, BD_TAG, 'Z', c->l_qseq+1, indelq);
bam_write1(tmp->out, b);
return 0;
}
int main(int argc , char *argv[])
{
FILE *fin , *fout ;
vcf_info vcf ;
faidx_t *fai ;
char buffer[32];
char *s = malloc(1024*sizeof(char));
if( argc != 4 ){
fprintf(stderr,"sv2vcf [ref.fa] [input] [output] \n");
return 1 ;
}
fai = fai_load(argv[1]);
if(!fai){
fprintf(stderr,"can't open the file:%s.fai\n",argv[1]);
return 1;
}
fin = fopen(argv[2],"r");
if(!fin){
fprintf(stderr,"can't open the file:%s\n", argv[2]);
return 1 ;
}
fout = fopen(argv[3],"w");
if(!fout){
fprintf(stderr,"can't open the file:%s\n", argv[3]);
return 1 ;
}
fprintf(fout,"#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILT\tINFO\n");
while(!feof(fin)){
char *tmp ;
int len ;
fscanf(fin,"%s\t",buffer);
if(!strcmp(buffer,"TRSdL")){
fscanf(fin,"%s\t",buffer);
fscanf(fin,"%d\t%d\n",&vcf.pos[0],&vcf.pos[1]);
int k ;
for( k = 0 ; k < 2 ; k++){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[k],vcf.pos[k]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[k] = *tmp ;
free(tmp);
}
fprintf(fout,"%s\t%d\tbnd_V\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],buffer,vcf.pos[1],vcf.mut[0]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],buffer,vcf.pos[0],vcf.mut[1]);
}else if(!strcmp(buffer,"TRSdR")){
fscanf(fin,"%s\t",buffer);
fscanf(fin,"%d\t%d\n",&vcf.pos[0],&vcf.pos[1]);
int k ;
for( k = 0 ; k < 2 ; k++){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[k],vcf.pos[k]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[k] = *tmp ;
free(tmp);
}
fprintf(fout,"%s\t%d\tbnd_V\t%c\t%s:%d]%c]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],buffer,vcf.pos[1],vcf.mut[0]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t%s:%d]%c]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],buffer,vcf.pos[0],vcf.mut[1]);
}else if((tmp = strstr(buffer,"TandemCNV"))){
tmp = tmp + 9 ;
vcf.type = atoi(tmp);
fscanf(fin,"%s\t",buffer);
if(vcf.type == 1 ) {
fscanf(fin,"%*[^\n]\n");
continue ;
}
fscanf(fin,"%d\t%d\n",&vcf.pos[0],&vcf.pos[1]);
int k ;
for( k = 0 ; k < 2 ; k++){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[k],vcf.pos[k]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[k] = *tmp ;
free(tmp);
}
if(vcf.type == 2){
fprintf(fout,"%s\t%d\tbnd_V\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],vcf.mut[1],buffer,vcf.pos[1]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],vcf.mut[1],buffer,vcf.pos[1]);
}else {
fprintf(fout,"%s\t%d\tbnd_V\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],buffer,vcf.pos[0],vcf.mut[0]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],buffer,vcf.pos[0],vcf.mut[0]);
}
}else if((tmp = strstr( buffer,"CNV"))){
tmp = tmp + 3;
vcf.type = atoi(tmp);
fscanf(fin,"%s\t",buffer);
fscanf(fin,"%d\t%d\t%d\n",&vcf.pos[0],&vcf.pos[1],&vcf.pos[2]);
if( vcf.type == 1 ){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[0],vcf.pos[0]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[0] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[2],vcf.pos[2]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[2] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[1],vcf.pos[1]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[1] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[2]+1,vcf.pos[2]+1);
tmp = fai_fetch(fai,s,&len);
vcf.mut[3] = *tmp ;
free(tmp);
fprintf(fout,"%s\t%d\tbnd_V\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],buffer,vcf.pos[2],vcf.mut[0]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[2],vcf.mut[2],vcf.mut[2],buffer,vcf.pos[0]);
fprintf(fout,"%s\t%d\tbnd_U\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_X;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],vcf.mut[1],buffer,vcf.pos[2]+1);
fprintf(fout,"%s\t%d\tbnd_X\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_U;EVENT=PRO\n",buffer,vcf.pos[2]+1,vcf.mut[3],buffer,vcf.pos[1],vcf.mut[3]);
}else if( vcf.type == 2 ){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[0],vcf.pos[0]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[0] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[2]+1,vcf.pos[2]+1);
tmp = fai_fetch(fai,s,&len);
vcf.mut[3] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[1],vcf.pos[1]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[1] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[2],vcf.pos[2]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[2] = *tmp ;
free(tmp);
fprintf(fout,"%s\t%d\tbnd_V\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],buffer,vcf.pos[2]+1,vcf.mut[0]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_U;EVENT=PRO\n",buffer,vcf.pos[2]+1,vcf.mut[3],buffer,vcf.pos[0],vcf.mut[3]);
fprintf(fout,"%s\t%d\tbnd_U\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_X;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],vcf.mut[1],buffer,vcf.pos[2]);
fprintf(fout,"%s\t%d\tbnd_X\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_U;EVENT=PRO\n",buffer,vcf.pos[2],vcf.mut[2],vcf.mut[2],buffer,vcf.pos[1]);
}else if(vcf.type == 3){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[0],vcf.pos[0]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[0] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[1],vcf.pos[1]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[1] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[0]+1,vcf.pos[0]+1);
tmp = fai_fetch(fai,s,&len);
vcf.mut[2] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[2],vcf.pos[2]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[3] = *tmp ;
free(tmp);
fprintf(fout,"%s\t%d\tbnd_V\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],vcf.mut[0],buffer,vcf.pos[1]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],buffer,vcf.pos[0],vcf.mut[1]);
fprintf(fout,"%s\t%d\tbnd_U\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_X;EVENT=PRO\n",buffer,vcf.pos[0]+1,vcf.mut[2],buffer,vcf.pos[2],vcf.mut[2]);
fprintf(fout,"%s\t%d\tbnd_X\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[2],vcf.mut[3],vcf.mut[3],buffer,vcf.pos[0]+1);
}else if(vcf.type == 4){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[0],vcf.pos[0]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[0] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[1],vcf.pos[1]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[1] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[0]+1,vcf.pos[0]+1);
tmp = fai_fetch(fai,s,&len);
vcf.mut[2] = *tmp ;
free(tmp);
sprintf(s,"%s:%d-%d",buffer,vcf.pos[2],vcf.pos[2]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[3] = *tmp ;
free(tmp);
fprintf(fout,"%s\t%d\tbnd_V\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],vcf.mut[0],buffer,vcf.pos[2]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[2],vcf.mut[3],vcf.mut[3],buffer,vcf.pos[0]);
fprintf(fout,"%s\t%d\tbnd_U\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_X;EVENT=PRO\n",buffer,vcf.pos[0]+1,vcf.mut[2],buffer,vcf.pos[1],vcf.mut[2]);
fprintf(fout,"%s\t%d\tbnd_X\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_U;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[3],buffer,vcf.pos[0]+1,vcf.mut[3]);
}
}else if((tmp =strstr(buffer,"TRS"))){
tmp = tmp + 3;
vcf.type = atoi(tmp);
fscanf(fin,"%s\t",buffer);
if( vcf.type == 2 || vcf.type == 3 ){
fscanf(fin,"%d\t%d\t%d\t%d\n",&vcf.pos[1],&vcf.pos[2],&vcf.pos[5],&vcf.pos[6]);
vcf.pos[0] = vcf.pos[1] - 1 ;
vcf.pos[3] = vcf.pos[2] + 1 ;
vcf.pos[4] = vcf.pos[5] - 1 ;
vcf.pos[7] = vcf.pos[6] + 1 ;
int k = 0 ;
for( k = 0 ; k < 8 ; k++){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[k],vcf.pos[k]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[k] = *tmp ;
free(tmp);
}
if(vcf.type == 2 ){
fprintf(fout,"%s\t%d\tbnd_V\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],vcf.mut[0],buffer,vcf.pos[6]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[6],vcf.mut[6],vcf.mut[6],buffer,vcf.pos[0]);
fprintf(fout,"%s\t%d\tbnd_U\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_X;EVENT=PRO\n",buffer,vcf.pos[3],vcf.mut[3],buffer,vcf.pos[5],vcf.mut[3]);
fprintf(fout,"%s\t%d\tbnd_X\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_U;EVENT=PRO\n",buffer,vcf.pos[5],vcf.mut[5],buffer,vcf.pos[3],vcf.mut[5]);
fprintf(fout,"%s\t%d\tbnd_E\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_D;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],buffer,vcf.pos[4],vcf.mut[1]);
fprintf(fout,"%s\t%d\tbnd_D\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_E;EVENT=PRO\n",buffer,vcf.pos[4],vcf.mut[4],vcf.mut[4],buffer,vcf.pos[1]);
fprintf(fout,"%s\t%d\tbnd_Z\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_H;EVENT=PRO\n",buffer,vcf.pos[2],vcf.mut[2],vcf.mut[2],buffer,vcf.pos[7]);
fprintf(fout,"%s\t%d\tbnd_H\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_Z;EVENT=PRO\n",buffer,vcf.pos[7],vcf.mut[7],buffer,vcf.pos[2],vcf.mut[7]);
}else {
fprintf(fout,"%s\t%d\tbnd_V\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],vcf.mut[0],buffer,vcf.pos[5]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_V;EVENT=PRO\n",buffer,vcf.pos[5],vcf.mut[5],buffer,vcf.pos[0],vcf.mut[5]);
fprintf(fout,"%s\t%d\tbnd_U\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_X;EVENT=PRO\n",buffer,vcf.pos[3],vcf.mut[3],buffer,vcf.pos[6],vcf.mut[3]);
fprintf(fout,"%s\t%d\tbnd_X\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_U;EVENT=PRO\n",buffer,vcf.pos[6],vcf.mut[6],vcf.mut[6],buffer,vcf.pos[3]);
fprintf(fout,"%s\t%d\tbnd_E\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_D;EVENT=PRO\n",buffer,vcf.pos[2],vcf.mut[2],vcf.mut[2],buffer,vcf.pos[4]);
fprintf(fout,"%s\t%d\tbnd_D\t%c\t%c]%s:%d]\t6\tPASS\tSVTYPE=BND;MATEID=bnd_E;EVENT=PRO\n",buffer,vcf.pos[4],vcf.mut[4],vcf.mut[4],buffer,vcf.pos[2]);
fprintf(fout,"%s\t%d\tbnd_Z\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_H;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],buffer,vcf.pos[7],vcf.mut[1]);
fprintf(fout,"%s\t%d\tbnd_H\t%c\t[%s:%d[%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_Z;EVENT=PRO\n",buffer,vcf.pos[7],vcf.mut[7],buffer,vcf.pos[1],vcf.mut[7]);
}
}else if(vcf.type == 1){
fscanf(fin,"%d\t%d\n",&vcf.pos[1],&vcf.pos[2]);
vcf.pos[0] = vcf.pos[1] - 1 ;
vcf.pos[3] = vcf.pos[2] + 1 ;
int k ;
for( k = 0 ; k < 4 ; k++){
sprintf(s,"%s:%d-%d",buffer,vcf.pos[k],vcf.pos[k]);
tmp = fai_fetch(fai,s,&len);
vcf.mut[k] = *tmp ;
free(tmp);
}
fprintf(fout,"%s\t%d\tbnd_V\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[0],vcf.mut[0],vcf.mut[0],buffer,vcf.pos[2]);
fprintf(fout,"%s\t%d\tbnd_W\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[2],vcf.mut[2],buffer,vcf.pos[0],vcf.mut[2]);
fprintf(fout,"%s\t%d\tbnd_U\t%c\t]%s:%d]%c\t6\tPASS\tSVTYPE=BND;MATEID=bnd_X;EVENT=PRO\n",buffer,vcf.pos[1],vcf.mut[1],buffer,vcf.pos[3],vcf.mut[1]);
fprintf(fout,"%s\t%d\tbnd_X\t%c\t%c[%s:%d[\t6\tPASS\tSVTYPE=BND;MATEID=bnd_W;EVENT=PRO\n",buffer,vcf.pos[3],vcf.mut[3],vcf.mut[3],buffer,vcf.pos[1]);
}else{
fscanf(fin,"%*[^\n]\n");
}
}else{
fscanf(fin,"%*[^\n]\n");
}
}
free(s);
fclose(fin);
fclose(fout);
return 0;
}
