Added error detection for cases such as a read's two mates having different names...

[rsem.git] / wiggle.cpp

diff --git a/wiggle.cpp b/wiggle.cpp
index 90b6f8b49d83ad5ddc45e3a9cd7d144d884c73ca..00dcce8d431edc2c38c0891d17b8085dbcbac25b 100644 (file)
--- a/wiggle.cpp
+++ b/wiggle.cpp
@@ -1,14 +1,27 @@
 #include <cstring>
 #include <cstdlib>
 #include <cassert>
+#include <iostream>
 
-#include "wiggle.h"
-
+#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) {
-    float w = bam_aux2f(bam_aux_get(b, "ZW"));
+    double 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);
     
@@ -34,27 +47,42 @@ void build_wiggles(const std::string& bam_filename,
                    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,
@@ -71,17 +99,19 @@ UCSCWiggleTrackWriter::~UCSCWiggleTrackWriter() {
 
 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++) {
-        if (wiggle.read_depth[i] > 0) {
+    for (size_t i = 0; i < wiggle.length; i++) {
+        if (wiggle.read_depth[i] >= 0.0095) {
             ep = i;
         }
         else {
             if (sp < ep) {
                 ++sp;
                 fprintf(fo, "fixedStep chrom=%s start=%d step=1\n", wiggle.name.c_str(), sp + 1);
-                for (int j = sp; j <= ep; j++) fprintf(fo, "%.7g\n", wiggle.read_depth[j]);
+                for (int j = sp; j <= ep; j++) fprintf(fo, "%.2f\n", wiggle.read_depth[j]);
             }
             sp = i;
         }
@@ -89,7 +119,7 @@ void UCSCWiggleTrackWriter::process(const Wiggle& wiggle) {
     if (sp < ep) {
         ++sp;
         fprintf(fo, "fixedStep chrom=%s start=%d step=1\n", wiggle.name.c_str(), sp + 1);
-        for (int j = sp; j <= ep; j++) fprintf(fo, "%.7g\n", wiggle.read_depth[j]);
+        for (int j = sp; j <= ep; j++) fprintf(fo, "%.2f\n", wiggle.read_depth[j]);
     }
 }
 
@@ -98,9 +128,13 @@ ReadDepthWriter::ReadDepthWriter(std::ostream& stream)
 }
 
 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];
     }