debian/changelog

+rapmap (0.15.0+dfsg-1) unstable; urgency=medium
+
+  * New upstream version
+  * debhelper-compat 12
+  * Standards-Version: 4.4.0
+  * Trim trailing whitespace.
+  * Set upstream metadata fields: Repository.
+  * -dev, -examples are Multi-Arch: foreign
+
+ -- Michael R. Crusoe <michael.crusoe@gmail.com>  Thu, 14 Nov 2019 14:45:15 +0100
+
 rapmap (0.14.1+dfsg-2) unstable; urgency=medium
 
   * only amd64 due to compiling variants for Intel SSE extensions

debian/compat

-11

debian/control

@@ -4,7 +4,7 @@ Uploaders: Andreas Tille <tille@debian.org>,
            Michael R. Crusoe <michael.crusoe@gmail.com>
 Section: science
 Priority: optional
-Build-Depends: debhelper (>= 11),
+Build-Depends: debhelper-compat (= 12),
                cmake,
                zlib1g-dev,
                libdivsufsort-dev,
@@ -15,8 +15,9 @@ Build-Depends: debhelper (>= 11),
                libspdlog-dev,
                libtclap-dev,
                pkg-config,
-               python3-markdown
-Standards-Version: 4.3.0
+               python3-markdown,
+               python3
+Standards-Version: 4.4.0
 Vcs-Browser: https://salsa.debian.org/med-team/rapmap
 Vcs-Git: https://salsa.debian.org/med-team/rapmap.git
 Homepage: https://github.com/COMBINE-lab/RapMap
@@ -52,6 +53,7 @@ Description: rapid sensitive and accurate DNA read mapping via quasi-mapping
 
 Package: rapmap-example-data
 Architecture: all
+Multi-Arch: foreign
 Depends: ${misc:Depends}
 Recommends: rapmap
 Description: example data for rapmap - rapid sensitive and accurate DNA read mapping
@@ -70,6 +72,7 @@ Description: example data for rapmap - rapid sensitive and accurate DNA read map
 
 Package: rapmap-dev
 Architecture: all
+Multi-Arch: foreign
 Depends: ${misc:Depends}
 Description: rapmap - rapid sensitive and accurate DNA read mapping (some headers)
  RapMap is a testing ground for ideas in quasi-mapping / (lightweight /

debian/rules

@@ -24,5 +24,3 @@ override_dh_auto_build:
 
 override_dh_clean:
 	dh_clean README.html sample_data/sample_quasi_index/ sample_data/sample_quasi_index_ph/ sample_data/sample_quasi_map.sam sample_data/sample_quasi_map_ph.sam
-
-

debian/upstream/metadata

 Reference:
 - Author: Avi Srivastava and Hirak Sarkar and Nitish Gupta and Rob Patro
-   Title: >
-    RapMap: a rapid, sensitive and accurate tool for mapping RNA-seq reads to
-    transcriptomes
+  Title: "RapMap: a rapid, sensitive and accurate tool for mapping RNA-seq reads to\
+    \ transcriptomes\n"
   Journal: Bioinformatics
   Year: 2016
   Volume: 32
@@ -31,3 +30,4 @@ Registry:
   Entry: NA
 - Name: conda:bioconda
   Entry: rapmap
+Repository: https://github.com/COMBINE-lab/RapMap

include/IndexHeader.hpp

@@ -54,6 +54,8 @@ class IndexHeader {
                 ar( cereal::make_nvp("NameHash", nameHash256_) );
                 ar( cereal::make_nvp("SeqHash512", seqHash512_) );
                 ar( cereal::make_nvp("NameHash512", nameHash512_) );
+                ar( cereal::make_nvp("DecoySeqHash", decoySeqHash256_) );
+                ar( cereal::make_nvp("DecoyNameHash", decoyNameHash256_) );
             }
 
         template <typename Archive>
@@ -69,6 +71,8 @@ class IndexHeader {
                 ar( cereal::make_nvp("NameHash", nameHash256_) );
                 ar( cereal::make_nvp("SeqHash512", seqHash512_) );
                 ar( cereal::make_nvp("NameHash512", nameHash512_) );
+                ar( cereal::make_nvp("DecoySeqHash", decoySeqHash256_) );
+                ar( cereal::make_nvp("DecoyNameHash", decoyNameHash256_) );
             } catch (const cereal::Exception& e) {
                 auto cerrLog = spdlog::get("stderrLog");
                 cerrLog->error("Encountered exception [{}] when loading index.", e.what());
@@ -91,10 +95,16 @@ class IndexHeader {
         std::string seqHash256() const { return seqHash256_; }
         std::string nameHash256() const { return nameHash256_; }
 
+        void setDecoySeqHash256(const std::string& seqHash) { decoySeqHash256_ = seqHash; }
+        void setDecoyNameHash256(const std::string& nameHash) { decoyNameHash256_ = nameHash; }
+        std::string decoySeqHash256() const { return decoySeqHash256_; }
+        std::string decoyNameHash256() const { return decoyNameHash256_; }
+
         void setSeqHash512(const std::string& seqHash) { seqHash512_ = seqHash; }
         void setNameHash512(const std::string& nameHash) { nameHash512_ = nameHash; }
         std::string seqHash512() const { return seqHash512_; }
         std::string nameHash512() const { return nameHash512_; }
+
     private:
         // The type of index we have
         IndexType type_;
@@ -117,6 +127,10 @@ class IndexHeader {
         std::string seqHash512_;
         // Hash of sequence names in txome
         std::string nameHash512_;
+        // Hash of the names of decoys
+        std::string decoyNameHash256_;
+        // Hash of the sequences of decoys
+        std::string decoySeqHash256_;
 };
 
 

src/RapMapSAIndexer.cpp

@@ -465,8 +465,11 @@ void indexTranscriptsSA(ParserT* parser,
   digestpp::sha256 nameHasher256;
   digestpp::sha512 seqHasher512;
   digestpp::sha512 nameHasher512;
-  //picosha2::hash256_one_by_one seqHasher; seqHasher.init();
-  //picosha2::hash256_one_by_one nameHasher; nameHasher.init();
+
+  digestpp::sha256 decoySeqHasher256;
+  digestpp::sha256 decoyNameHasher256;
+
+
 
   bool firstRecord{true};
   bool hasGencodeSep = (sepStr.find('|') != std::string::npos);
@@ -536,9 +539,6 @@ void indexTranscriptsSA(ParserT* parser,
                            [](const char a) -> bool { return !(isprint(a)); }),
             readStr.end());
 
-        seqHasher256.absorb(readStr.begin(), readStr.end());
-        seqHasher512.absorb(readStr.begin(), readStr.end());
-
         uint32_t readLen = readStr.size();
         uint32_t completeLen = readLen;
 
@@ -573,6 +573,14 @@ void indexTranscriptsSA(ParserT* parser,
           //throw std::logic_error("Input fasta file contained out-of-order decoy targets.");
         }
 
+        // If this was a decoy, add it to the decoy hash
+        if (isDecoy) {
+          decoySeqHasher256.absorb(readStr.begin(), readStr.end());
+        } else { // otherwise the ref hash
+          seqHasher256.absorb(readStr.begin(), readStr.end());
+          seqHasher512.absorb(readStr.begin(), readStr.end());
+        }
+
         // First, replace non ATCG nucleotides
         for (size_t b = 0; b < readLen; ++b) {
           readStr[b] = ::toupper(readStr[b]);
@@ -656,8 +664,12 @@ void indexTranscriptsSA(ParserT* parser,
 
           // If there was no collision, then add the transcript
           transcriptNames.emplace_back(processedName);
+          if (isDecoy) {
+            decoyNameHasher256.absorb(processedName.begin(), processedName.end());
+          } else {
             nameHasher256.absorb(processedName.begin(), processedName.end());
             nameHasher512.absorb(processedName.begin(), processedName.end());
+          }
 
           // The position at which this transcript starts
           transcriptStarts.push_back(currIndex);
@@ -869,10 +881,15 @@ void indexTranscriptsSA(ParserT* parser,
   std::string nameHash256 = nameHasher256.hexdigest();
   std::string seqHash512 = seqHasher512.hexdigest();
   std::string nameHash512 = nameHasher512.hexdigest();
+  std::string decoySeqHash256 = decoySeqHasher256.hexdigest();
+  std::string decoyNameHash256 = decoyNameHasher256.hexdigest();
+
   header.setSeqHash256(seqHash256);
   header.setNameHash256(nameHash256);
   header.setSeqHash512(seqHash512);
   header.setNameHash512(nameHash512);
+  header.setDecoySeqHash256(decoySeqHash256);
+  header.setDecoyNameHash256(decoyNameHash256);
   //std::string seqHash;
   //std::string nameHash;
   //picosha2::get_hash_hex_string(seqHasher, seqHash);