#include <cstring>
#include <cstdlib>
#include <cassert>
+#include <iostream>
#include <stdint.h>
#include "sam/bam.h"
#include "sam/sam.h"
+#include "utils.h"
#include "wiggle.h"
+bool no_fractional_weight = false;
+
void add_bam_record_to_wiggle(const bam1_t *b, Wiggle& wiggle) {
- uint8_t *p_tag = bam_aux_get(b, "ZW");
- float w = (p_tag != NULL ? bam_aux2f(p_tag) : 1.0);
+ float w;
+
+ if (no_fractional_weight) w = 1.0;
+ else {
+ uint8_t *p_tag = bam_aux_get(b, "ZW");
+ if (p_tag == NULL) return;
+ w = bam_aux2f(p_tag);
+ }
+
int pos = b->core.pos;
uint32_t *p = bam1_cigar(b);
WiggleProcessor& processor) {
samfile_t *bam_in = samopen(bam_filename.c_str(), "rb", NULL);
if (bam_in == 0) { fprintf(stderr, "Cannot open %s!\n", bam_filename.c_str()); exit(-1); }
- //assert(bam_in != 0);
- int cur_tid = -1; //current tid;
- int cnt = 0;
- bam1_t *b = bam_init1();
- Wiggle wiggle;
+ bam_header_t *header = bam_in->header;
+ bool *used = new bool[header->n_targets];
+ memset(used, 0, sizeof(bool) * header->n_targets);
+
+ int cur_tid = -1; //current tid;
+ HIT_INT_TYPE cnt = 0;
+ bam1_t *b = bam_init1();
+ Wiggle wiggle;
while (samread(bam_in, b) >= 0) {
if (b->core.flag & 0x0004) continue;
if (b->core.tid != cur_tid) {
- if (cur_tid >= 0) processor.process(wiggle);
+ if (cur_tid >= 0) { used[cur_tid] = true; processor.process(wiggle); }
cur_tid = b->core.tid;
- wiggle.name = bam_in->header->target_name[cur_tid];
- wiggle.read_depth.assign(bam_in->header->target_len[cur_tid], 0.0);
+ wiggle.name = header->target_name[cur_tid];
+ wiggle.length = header->target_len[cur_tid];
+ wiggle.read_depth.assign(wiggle.length, 0.0);
}
- add_bam_record_to_wiggle(b, wiggle);
+ add_bam_record_to_wiggle(b, wiggle);
++cnt;
- if (cnt % 1000000 == 0) fprintf(stderr, "%d FIN\n", cnt);
+ if (cnt % 1000000 == 0) std::cout<< cnt<< std::endl;
}
- if (cur_tid >= 0) processor.process(wiggle);
+ if (cur_tid >= 0) { used[cur_tid] = true; processor.process(wiggle); }
+
+ for (int32_t i = 0; i < header->n_targets; i++)
+ if (!used[i]) {
+ wiggle.name = header->target_name[i];
+ wiggle.length = header->target_len[i];
+ wiggle.read_depth.clear();
+ processor.process(wiggle);
+ }
samclose(bam_in);
bam_destroy1(b);
+ delete[] used;
}
UCSCWiggleTrackWriter::UCSCWiggleTrackWriter(const std::string& output_filename,
void UCSCWiggleTrackWriter::process(const Wiggle& wiggle) {
int sp, ep;
+
+ if (wiggle.read_depth.empty()) return;
sp = ep = -1;
- for (size_t i = 0; i < wiggle.read_depth.size(); i++) {
+ for (size_t i = 0; i < wiggle.length; i++) {
if (wiggle.read_depth[i] > 0) {
ep = i;
}
}
void ReadDepthWriter::process(const Wiggle& wiggle) {
+
stream_ << wiggle.name << '\t'
- << wiggle.read_depth.size() << '\t';
- for (size_t i = 0; i < wiggle.read_depth.size(); ++i) {
+ << wiggle.length << '\t';
+
+ if (wiggle.read_depth.empty()) { stream_ << "NA\n"; return; }
+
+ for (size_t i = 0; i < wiggle.length; ++i) {
if (i > 0) stream_ << ' ';
stream_ << wiggle.read_depth[i];
}