int64_t bns_fasta2bntseq(gzFile fp_fa, const char *prefix, int for_only)
{
extern void seq_reverse(int len, ubyte_t *seq, int is_comp); // in bwaseqio.c
kseq_t *seq;
char name[1024];
bntseq_t *bns;
uint8_t *pac = 0;
int32_t m_seqs, m_holes;
int64_t ret = -1, m_pac, l;
bntamb1_t *q;
FILE *fp;
// initialization
seq = kseq_init(fp_fa);
bns = (bntseq_t*)calloc(1, sizeof(bntseq_t));
bns->seed = 11; // fixed seed for random generator
srand48(bns->seed);
m_seqs = m_holes = 8; m_pac = 0x10000;
bns->anns = (bntann1_t*)calloc(m_seqs, sizeof(bntann1_t));
bns->ambs = (bntamb1_t*)calloc(m_holes, sizeof(bntamb1_t));
pac = calloc(m_pac/4, 1);
q = bns->ambs;
strcpy(name, prefix); strcat(name, ".pac");
fp = xopen(name, "wb");
// read sequences
while (kseq_read(seq) >= 0) pac = add1(seq, bns, pac, &m_pac, &m_seqs, &m_holes, &q);
if (!for_only) { // add the reverse complemented sequence
m_pac = (bns->l_pac * 2 + 3) / 4 * 4;
pac = realloc(pac, m_pac/4);
memset(pac + (bns->l_pac+3)/4, 0, (m_pac - (bns->l_pac+3)/4*4) / 4);
for (l = bns->l_pac - 1; l >= 0; --l, ++bns->l_pac)
_set_pac(pac, bns->l_pac, 3-_get_pac(pac, l));
}
ret = bns->l_pac;
{ // finalize .pac file
ubyte_t ct;
err_fwrite(pac, 1, (bns->l_pac>>2) + ((bns->l_pac&3) == 0? 0 : 1), fp);
// the following codes make the pac file size always (l_pac/4+1+1)
if (bns->l_pac % 4 == 0) {
ct = 0;
err_fwrite(&ct, 1, 1, fp);
}
ct = bns->l_pac % 4;
err_fwrite(&ct, 1, 1, fp);
// close .pac file
err_fflush(fp);
err_fclose(fp);
}
bns_dump(bns, prefix);
bns_destroy(bns);
kseq_destroy(seq);
free(pac);
return ret;
}
static uint8_t *bis_add1(const kseq_t *seq, bntseq_t *bns, uint8_t *pac, int64_t *m_pac, int *m_seqs, int *m_holes, bntamb1_t **q)
{
bntann1_t *p;
int lasts;
uint32_t i;
if (bns->n_seqs == *m_seqs) {
*m_seqs <<= 1;
bns->anns = (bntann1_t*)realloc(bns->anns, *m_seqs * sizeof(bntann1_t));
}
p = bns->anns + bns->n_seqs;
p->name = strdup((char*)seq->name.s);
p->anno = seq->comment.s? strdup((char*)seq->comment.s) : strdup("(null)");
p->gi = 0; p->len = seq->seq.l;
p->offset = (bns->n_seqs == 0)? 0 : (p-1)->offset + (p-1)->len;
p->n_ambs = 0;
for (i = lasts = 0; i < seq->seq.l; ++i) {
int c = nst_nt4_table[(int)seq->seq.s[i]];
if (c >= 4) { // N
if (lasts == seq->seq.s[i]) { // contiguous N
++(*q)->len;
} else {
if (bns->n_holes == *m_holes) {
(*m_holes) <<= 1;
bns->ambs = (bntamb1_t*)realloc(bns->ambs, (*m_holes) * sizeof(bntamb1_t));
}
*q = bns->ambs + bns->n_holes;
(*q)->len = 1;
(*q)->offset = p->offset + i;
(*q)->amb = seq->seq.s[i];
++p->n_ambs;
++bns->n_holes;
}
}
lasts = seq->seq.s[i];
{ // fill buffer
if (c >= 4) c = lrand48()&3;
if (bns->l_pac == *m_pac) { // double the pac size
*m_pac <<= 1;
pac = realloc(pac, *m_pac/4);
memset(pac + bns->l_pac/4, 0, (*m_pac - bns->l_pac)/4);
}
_set_pac(pac, bns->l_pac, c);
++bns->l_pac;
}
}
++bns->n_seqs;
return pac;
}
int64_t bis_bns_fasta2bntseq(gzFile fp_fa, const char *prefix, uint8_t parent) {
extern void seq_reverse(int len, ubyte_t *seq, int is_comp); // in bwaseqio.c
kseq_t *seq;
char name[1024];
bntseq_t *bns;
uint8_t *pac = 0, *_pac = 0;
int32_t m_seqs, m_holes;
int64_t ret = -1, m_pac;
bntamb1_t *q;
FILE *fp;
// initialization
gzseek(fp_fa, 0, SEEK_SET);
seq = kseq_init(fp_fa);
bns = (bntseq_t*)calloc(1, sizeof(bntseq_t));
bns->seed = 11; // fixed seed for random generator
srand48(bns->seed);
m_seqs = m_holes = 8; m_pac = 0x10000;
bns->anns = (bntann1_t*)calloc(m_seqs, sizeof(bntann1_t));
bns->ambs = (bntamb1_t*)calloc(m_holes, sizeof(bntamb1_t));
_pac = calloc(m_pac/4, 1);
q = bns->ambs;
if (parent) {
strcpy(name, prefix); strcat(name, ".par.pac");
} else {
strcpy(name, prefix); strcat(name, ".dau.pac");
}
fp = xopen(name, "wb");
// read sequences
while (kseq_read(seq) >= 0) {
_pac = bis_add1(seq, bns, _pac, &m_pac, &m_seqs, &m_holes, &q);
}
/* kseq_rewind(seq); */
/* gzseek(seq->f->f, 0, SEEK_SET); */
/* fprintf(stderr, "foward end\n"); */
/* fflush(stderr); */
/* while (kseq_read(seq) >= 0) { */
/* if (parent) nt256char_rev_ip(seq->seq.s, seq->seq.l); */
/* pac = bis_add1(seq, bns, pac, &m_pac, &m_seqs, &m_holes, &q, parent, 1); */
/* } */
int64_t l,k;
m_pac = (bns->l_pac*2+3)/4*4; /* in bit */
pac = calloc(m_pac/4,sizeof(uint8_t));
for (l=0; l<bns->l_pac; ++l) {
uint8_t c = _get_pac(_pac,l);
if (parent && c == 1) c = 3;
if (!parent && c == 2) c = 0;
_set_pac(pac, l, c);
}
for (k=bns->l_pac-1; k>=0; --k,++l) {
uint8_t c = 3-_get_pac(_pac,k);
if (parent && c == 1) c = 3;
if (!parent && c == 2) c = 0;
_set_pac(pac, l, c);
}
free(_pac);
/* int64_t l; */
/* fprintf(stderr, "reverse end\n"); */
/* fflush(stderr); */
/* if (!for_only) { // add the reverse complemented sequence */
/* m_pac = (bns->l_pac * 2 + 3) / 4 * 4; */
/* pac = realloc(pac, m_pac/4); */
/* memset(pac + (bns->l_pac+3)/4, 0, (m_pac - (bns->l_pac+3)/4*4) / 4); */
/* for (l = bns->l_pac - 1; l >= 0; --l, ++bns->l_pac) */
/* _set_pac(pac, bns->l_pac, 3-_get_pac(pac, l)); */
/* } */
assert(bns->l_pac<<1 == l);
{ // finalize .pac file
ubyte_t ct;
err_fwrite(pac, 1, (l>>2) + ((l&3) == 0? 0 : 1), fp);
// the following codes make the pac file size always (l_pac/4+1+1)
if (l % 4 == 0) {
ct = 0;
err_fwrite(&ct, 1, 1, fp);
}
ct = l % 4;
err_fwrite(&ct, 1, 1, fp);
// close .pac file
err_fflush(fp);
err_fclose(fp);
}
if (parent) bis_bns_dump(bns, prefix);
bns_destroy(bns);
kseq_destroy(seq);
free(pac);
return l;
}
int64_t R_bns_fasta2bntseq(gzFile fp_fa, const char *prefix)
{
// extern void seq_reverse(int len, ubyte_t *seq, int is_comp); // in bwaseqio.c
kseq_t *seq;
char name[1024];
bntseq_t *bns;
uint8_t *pac = 0;
uint8_t *reverse_pac = NULL;
int32_t m_seqs, m_holes;
int64_t ret = -1, m_pac, l, m_r_pac;
bntamb1_t *q;
FILE *fp, *fp_r;
int i;
// initialization
seq = kseq_init(fp_fa);
bns = (bntseq_t*)calloc(1, sizeof(bntseq_t));
bns->seed = 11; // fixed seed for random generator
srand48(bns->seed);
m_seqs = m_holes = 8; m_pac = 0x10000;
bns->anns = (bntann1_t*)calloc(m_seqs, sizeof(bntann1_t));
bns->ambs = (bntamb1_t*)calloc(m_holes, sizeof(bntamb1_t));
pac = calloc(m_pac/NT_PER_BYTE, 1);
q = bns->ambs;
strcpy(name, prefix); strcat(name, ".pac");
fp = xopen(name, "wb");
memset(name, '\0', 1024);strcpy(name, prefix); strcat(name, ".rpac");
fp_r = xopen(name, "wb");
// read sequences
while (kseq_read(seq) >= 0) pac = R_add1(seq, bns, pac, &m_pac, &m_seqs, &m_holes, &q);
// add the reverse complemented sequence
ret = bns->l_pac;
fprintf(stderr, "[R_bns_fasta2bntseq]:reverse_pac!\n");
m_r_pac = (bns->l_pac +NT_PER_BYTE-1)/NT_PER_BYTE *NT_PER_BYTE;
reverse_pac = calloc(m_r_pac/NT_PER_BYTE, sizeof(uint8_t));
for(l = bns->l_pac-1, i =0; l>=0, i < bns->l_pac; --l, ++i)
{
_set_pac(reverse_pac, i, _get_pac(pac, l));
}
ubyte_t ct;
{ // finalize .pac and file
fwrite(pac, 1, (bns->l_pac>>1) + ((bns->l_pac&1) == 0? 0 : 1), fp);
// the following codes make the pac file size always (l_pac/4+1+1)
if (bns->l_pac % NT_PER_BYTE == 0) {
ct = 0;
fwrite(&ct, 1, 1, fp);
}
ct = bns->l_pac % NT_PER_BYTE;
fwrite(&ct, 1, 1, fp);
// close .pac file
fclose(fp);
}
{ // finalize .rpac and file
fwrite(reverse_pac, 1, (bns->l_pac>>1) + ((bns->l_pac&1) == 0? 0 : 1), fp_r);
// the following codes make the pac file size always (l_pac/4+1+1)
if (bns->l_pac % NT_PER_BYTE == 0) {
ct = 0;
fwrite(&ct, 1, 1, fp_r);
}
ct = bns->l_pac % NT_PER_BYTE;
fwrite(&ct, 1, 1, fp_r);
// close .rpac file
fclose(fp_r);
}
bns_dump(bns, prefix);
bns_destroy(bns);
kseq_destroy(seq);
free(pac);
free(reverse_pac);
return ret;
}