inline void
tmap_seqs_io_destroy(tmap_seqs_io_t *io)
{
int32_t i;
for(i=0;i<io->n;i++) {
tmap_seq_io_destroy(io->seqios[i]);
}
free(io->seqios);
free(io);
}
uint64_t
tmap_refseq_fasta2pac(const char *fn_fasta, int32_t compression)
{
tmap_file_t *fp_pac = NULL, *fp_anno = NULL;
tmap_seq_io_t *seqio = NULL;
tmap_seq_t *seq = NULL;
tmap_refseq_t *refseq = NULL;
char *fn_pac = NULL, *fn_anno = NULL;
uint8_t buffer[TMAP_REFSEQ_BUFFER_SIZE];
int32_t i, j, l, buffer_length;
uint32_t num_IUPAC_found= 0, amb_bases_mem = 0;
uint8_t x = 0;
uint64_t ref_len;
tmap_progress_print("packing the reference FASTA");
refseq = tmap_calloc(1, sizeof(tmap_refseq_t), "refseq");
refseq->version_id = TMAP_VERSION_ID;
refseq->package_version = tmap_string_clone2(PACKAGE_VERSION);
refseq->seq = buffer; // IMPORTANT: must nullify later
refseq->annos = NULL;
refseq->num_annos = 0;
refseq->len = 0;
refseq->is_rev = 0;
refseq->is_shm = 0;
memset(buffer, 0, TMAP_REFSEQ_BUFFER_SIZE);
buffer_length = 0;
// input files
seqio = tmap_seq_io_init(fn_fasta, TMAP_SEQ_TYPE_FQ, 0, compression);
seq = tmap_seq_init(TMAP_SEQ_TYPE_FQ);
// output files
fn_pac = tmap_get_file_name(fn_fasta, TMAP_PAC_FILE);
fp_pac = tmap_file_fopen(fn_pac, "wb", TMAP_PAC_COMPRESSION);
// read in sequences
while(0 <= (l = tmap_seq_io_read(seqio, seq))) {
tmap_anno_t *anno = NULL;
tmap_progress_print2("packing contig [%s:1-%d]", seq->data.fq->name->s, l);
refseq->num_annos++;
refseq->annos = tmap_realloc(refseq->annos, sizeof(tmap_anno_t)*refseq->num_annos, "refseq->annos");
anno = &refseq->annos[refseq->num_annos-1];
anno->name = tmap_string_clone(seq->data.fq->name);
anno->len = l;
anno->offset = (1 == refseq->num_annos) ? 0 : refseq->annos[refseq->num_annos-2].offset + refseq->annos[refseq->num_annos-2].len;
anno->amb_positions_start = NULL;
anno->amb_positions_end = NULL;
anno->amb_bases = NULL;
anno->num_amb = 0;
amb_bases_mem = 0;
// fill the buffer
for(i=0;i<l;i++) {
uint8_t c = tmap_nt_char_to_int[(int)seq->data.fq->seq->s[i]];
// handle IUPAC codes
if(4 <= c) {
int32_t k;
// warn users about IUPAC codes
if(0 == num_IUPAC_found) {
tmap_error("IUPAC codes were found and will be converted to non-matching DNA bases", Warn, OutOfRange);
for(j=4;j<15;j++) {
c = tmap_iupac_char_to_bit_string[(int)tmap_iupac_int_to_char[j]];
// get the lexicographically smallest base not compatible with this code
for(k=0;k<4;k++) {
if(!(c & (0x1 << k))) {
break;
}
}
tmap_progress_print2("IUPAC code %c will be converted to %c", tmap_iupac_int_to_char[j], "ACGTN"[k & 3]);
}
}
num_IUPAC_found++;
// change it to a mismatched base than the IUPAC code
c = tmap_iupac_char_to_bit_string[(int)seq->data.fq->seq->s[i]];
// store IUPAC bases
if(amb_bases_mem <= anno->num_amb) { // allocate more memory if necessary
amb_bases_mem = anno->num_amb + 1;
tmap_roundup32(amb_bases_mem);
anno->amb_positions_start = tmap_realloc(anno->amb_positions_start, sizeof(uint32_t) * amb_bases_mem, "anno->amb_positions_start");
anno->amb_positions_end = tmap_realloc(anno->amb_positions_end, sizeof(uint32_t) * amb_bases_mem, "anno->amb_positions_end");
anno->amb_bases = tmap_realloc(anno->amb_bases, sizeof(uint8_t) * amb_bases_mem, "anno->amb_bases");
}
// encode stretches of the same base
if(0 < anno->num_amb
&& anno->amb_positions_end[anno->num_amb-1] == i
&& anno->amb_bases[anno->num_amb-1] == tmap_iupac_char_to_int[(int)seq->data.fq->seq->s[i]]) {
anno->amb_positions_end[anno->num_amb-1]++; // expand the range
}
else {
// new ambiguous base and range
anno->num_amb++;
anno->amb_positions_start[anno->num_amb-1] = i+1; // one-based
anno->amb_positions_end[anno->num_amb-1] = i+1; // one-based
anno->amb_bases[anno->num_amb-1] = tmap_iupac_char_to_int[(int)seq->data.fq->seq->s[i]];
}
// get the lexicographically smallest base not compatible with
// this code
for(j=0;j<4;j++) {
if(!(c & (0x1 << j))) {
break;
}
}
c = j & 3; // Note: Ns will go to As
}
if(3 < c) {
tmap_error("bug encountered", Exit, OutOfRange);
}
if(buffer_length == (TMAP_REFSEQ_BUFFER_SIZE << 2)) { // 2-bit
if(tmap_refseq_seq_memory(buffer_length) != tmap_file_fwrite(buffer, sizeof(uint8_t), tmap_refseq_seq_memory(buffer_length), fp_pac)) {
tmap_error(fn_pac, Exit, WriteFileError);
}
memset(buffer, 0, TMAP_REFSEQ_BUFFER_SIZE);
buffer_length = 0;
}
tmap_refseq_seq_store_i(refseq, buffer_length, c);
buffer_length++;
}
refseq->len += l;
// re-size the amibiguous bases
if(anno->num_amb < amb_bases_mem) {
amb_bases_mem = anno->num_amb;
anno->amb_positions_start = tmap_realloc(anno->amb_positions_start, sizeof(uint32_t) * amb_bases_mem, "anno->amb_positions_start");
anno->amb_positions_end = tmap_realloc(anno->amb_positions_end, sizeof(uint32_t) * amb_bases_mem, "anno->amb_positions_end");
anno->amb_bases = tmap_realloc(anno->amb_bases, sizeof(uint8_t) * amb_bases_mem, "anno->amb_bases");
}
}
// write out the buffer
if(tmap_refseq_seq_memory(buffer_length) != tmap_file_fwrite(buffer, sizeof(uint8_t), tmap_refseq_seq_memory(buffer_length), fp_pac)) {
tmap_error(fn_pac, Exit, WriteFileError);
}
if(refseq->len % 4 == 0) { // add an extra byte if we completely filled all bits
if(1 != tmap_file_fwrite(&x, sizeof(uint8_t), 1, fp_pac)) {
tmap_error(fn_pac, Exit, WriteFileError);
}
}
// store number of unused bits at the last byte
x = refseq->len % 4;
if(1 != tmap_file_fwrite(&x, sizeof(uint8_t), 1, fp_pac)) {
tmap_error(fn_pac, Exit, WriteFileError);
}
refseq->seq = NULL; // IMPORTANT: nullify this
ref_len = refseq->len; // save for return
tmap_progress_print2("total genome length [%u]", refseq->len);
if(0 < num_IUPAC_found) {
if(1 == num_IUPAC_found) {
tmap_progress_print("%u IUPAC base was found and converted to a DNA base", num_IUPAC_found);
}
else {
tmap_progress_print("%u IUPAC bases were found and converted to DNA bases", num_IUPAC_found);
}
}
// write annotation file
fn_anno = tmap_get_file_name(fn_fasta, TMAP_ANNO_FILE);
fp_anno = tmap_file_fopen(fn_anno, "wb", TMAP_ANNO_COMPRESSION);
tmap_refseq_write_anno(fp_anno, refseq);
// close files
tmap_file_fclose(fp_pac);
tmap_file_fclose(fp_anno);
// check sequence name uniqueness
for(i=0;i<refseq->num_annos;i++) {
for(j=i+1;j<refseq->num_annos;j++) {
if(0 == strcmp(refseq->annos[i].name->s, refseq->annos[j].name->s)) {
tmap_file_fprintf(tmap_file_stderr, "Contigs have the same name: #%d [%s] and #%d [%s]\n",
i+1, refseq->annos[i].name->s,
j+1, refseq->annos[j].name->s);
tmap_error("Contig names must be unique", Exit, OutOfRange);
}
}
}
tmap_refseq_destroy(refseq);
tmap_seq_io_destroy(seqio);
tmap_seq_destroy(seq);
free(fn_pac);
free(fn_anno);
tmap_progress_print2("packed the reference FASTA");
tmap_refseq_pac2revpac(fn_fasta);
return ref_len;
}