Updates to BUS format, fixes htslib config mistake

pachterlab · Nov 16, 2018 · 95dc71a · 95dc71a
1 parent b6f6dc5
commit 95dc71a
Show file tree

Hide file tree

Showing 5 changed files with 33 additions and 8 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -40,7 +40,7 @@ ExternalProject_Add(htslib
     PREFIX ${PROJECT_SOURCE_DIR}/ext/htslib
     SOURCE_DIR ${PROJECT_SOURCE_DIR}/ext/htslib
     BUILD_IN_SOURCE 1
-    CONFIGURE_COMMAND ${PROJECT_SOURCE_DIR}/ext/htslib/configure 
+    CONFIGURE_COMMAND autoheader && autoconf && ${PROJECT_SOURCE_DIR}/ext/htslib/configure 
         --prefix=${PREFIX} --disable-bz2 --disable-lzma --disable-libcurl
     BUILD_COMMAND make lib-static
     INSTALL_COMMAND ""

diff --git a/src/BUSData.h b/src/BUSData.h
@@ -5,6 +5,8 @@
 #include <vector>
 #include <stdint.h>
 
+const uint32_t BUSFORMAT_VERSION = 1;
+
 struct BUSTranscript {
   std::string name;
   uint32_t transcriptLength;
@@ -13,7 +15,7 @@ struct BUSTranscript {
 
 
 struct BUSHeader {
-  std::string text; 
+  std::string text;
   std::vector<BUSTranscript> transcripts;
   std::vector<std::vector<int32_t>> ecs;
 };

diff --git a/src/BUSTools.cpp b/src/BUSTools.cpp
@@ -3,8 +3,14 @@
 #include "BUSData.h"
 
 void writeBUSHeader(std::ofstream &out, int bclen, int umilen) {
+  out.write("BUS\0", 4);
+  out.write((char*)(&BUSFORMAT_VERSION), sizeof(BUSFORMAT_VERSION));
   out.write((char*)(&bclen), sizeof(bclen));
   out.write((char*)(&umilen), sizeof(umilen));
+  std::string header_text = "BUS file produced by kallisto";
+  uint32_t len = header_text.size();
+  out.write((char*)(&len),sizeof(len));  
+  out.write(header_text.c_str(), len);
 }
 
 void writeBUSData(std::ofstream &out, const std::vector<BUSData> &bv) {

diff --git a/src/ProcessReads.cpp b/src/ProcessReads.cpp
@@ -1338,21 +1338,31 @@ void BUSProcessor::processBuffer() {
 
     // copy the umi
     int umilen = (busopt.umi.start == busopt.umi.stop) ? l[busopt.umi.fileno] - busopt.umi.start : busopt.umi.stop - busopt.umi.start;
+    if (l[busopt.umi.fileno] < busopt.umi.start + umilen) {
+      continue; // too short
+    }
     memcpy(umi, s[busopt.umi.fileno] + busopt.umi.start, umilen);
     umi[umilen] = 0;
     if (umilen >= 0 && umilen <= 32) {
       umi_len[umilen]++;
     }
 
-
-    // TODO handle concatenated barcodes
+
     auto &bcc = busopt.bc[0];
     int blen = 0;
+    bool bad_bc = false;
     for (auto &bcc : busopt.bc) {
       int bclen = (bcc.start == bcc.stop) ? l[bcc.fileno] - bcc.start : bcc.stop - bcc.start;
+      if (l[bcc.fileno] < bcc.start + bclen) {
+        bad_bc = true;
+        break;
+      }
       memcpy(bc+blen, s[bcc.fileno] + bcc.start, bclen);
       blen += bclen;
     }
+    if (bad_bc) {
+      continue;
+    }
     bc[blen] = 0;
 
     if (blen >= 0 && blen <= 32) {

diff --git a/src/main.cpp b/src/main.cpp
@@ -1644,12 +1644,11 @@ int main(int argc, char *argv[]) {
         }
 
 
-        if (write) {
-          //std::cout << bclen << "\t" << umilen << endl;
+        if (write) {          
           std::FILE* fp = std::fopen((opt.output + "/output.bus").c_str(), "r+b");
           if (fp != nullptr) {
-            std::fseek(fp,0,SEEK_SET);
-            //write to int values
+            std::fseek(fp,8,SEEK_SET); // skip magic string and version
+            // write to uint32_t values
             std::fwrite(&bclen, sizeof(bclen),1,fp);
             std::fwrite(&umilen, sizeof(umilen),1,fp);
             std::fclose(fp);
@@ -1660,6 +1659,14 @@ int main(int argc, char *argv[]) {
 
         writeECList(opt.output + "/matrix.ec", index);
 
+        // write transcript names
+        std::ofstream transout_f((opt.output + "/transcripts.txt"));
+        for (const auto &t : index.target_names_) {
+          transout_f << t << "\n";
+        }
+        transout_f.close();
+
+
         // gather stats
         num_unique = 0;
         for (int i = 0; i < index.num_trans; i++) {