Transfer code from internal repository

Illumina · Jan 17, 2019 · 5818874 · 5818874
1 parent a0ad4f4
commit 5818874
Show file tree

Hide file tree

Showing 239 changed files with 2,424 additions and 2,095 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -68,7 +68,6 @@ find_package(Boost 1.4 REQUIRED COMPONENTS program_options filesystem regex date
 include_directories(${CMAKE_CURRENT_SOURCE_DIR})
 include_directories(SYSTEM ${Boost_INCLUDE_DIR})
 include_directories(${CMAKE_BINARY_DIR}/thirdparty/htslib/include)
-include_directories(thirdparty/graph-tools-GT-506/include)
 
 add_subdirectory(thirdparty/graph-tools-master)
 

diff --git a/COPYRIGHT.txt b/COPYRIGHT.txt
diff --git a/LICENSE.txt b/LICENSE.txt
diff --git a/README.md b/README.md
diff --git a/alignment/AlignmentFilters.cpp b/alignment/AlignmentFilters.cpp
@@ -32,6 +32,8 @@
 
 using graphtools::GraphAlignment;
 using graphtools::NodeId;
+using graphtools::Operation;
+using graphtools::OperationType;
 using graphtools::Path;
 using std::list;
 using std::string;
@@ -110,4 +112,28 @@ bool checkIfDownstreamAlignmentIsGood(NodeId nodeId, GraphAlignment alignment)
     return score >= kScoreCutoff;
 }
 
+bool checkIfPassesAlignmentFilters(const GraphAlignment& alignment)
+{
+    const Operation& firstOperation = alignment.alignments().front().operations().front();
+    const int frontSoftclipLen = firstOperation.type() == OperationType::kSoftclip ? firstOperation.queryLength() : 0;
+
+    const Operation& lastOperation = alignment.alignments().back().operations().back();
+    const int backSoftclipLen = lastOperation.type() == OperationType::kSoftclip ? lastOperation.queryLength() : 0;
+
+    const int clippedQueryLength = alignment.queryLength() - frontSoftclipLen - backSoftclipLen;
+    const int referenceLength = alignment.referenceLength();
+
+    const int percentQueryMatches = (100 * alignment.numMatches()) / clippedQueryLength;
+    const int percentReferenceMatches = (100 * alignment.numMatches()) / referenceLength;
+
+    if (percentQueryMatches >= 80 && percentReferenceMatches >= 80)
+    {
+        return true;
+    }
+    else
+    {
+        return false;
+    }
+}
+
 }
diff --git a/alignment/AlignmentFilters.hh b/alignment/AlignmentFilters.hh
@@ -50,4 +50,6 @@ bool checkIfUpstreamAlignmentIsGood(graphtools::NodeId nodeId, graphtools::Graph
 // Checks if alignment downstream of a given node is high quality
 bool checkIfDownstreamAlignmentIsGood(graphtools::NodeId nodeId, graphtools::GraphAlignment alignment);
 
+bool checkIfPassesAlignmentFilters(const graphtools::GraphAlignment& alignment);
+
 }
diff --git a/alignment/AlignmentTweakers.cpp b/alignment/AlignmentTweakers.cpp
diff --git a/alignment/AlignmentTweakers.hh b/alignment/AlignmentTweakers.hh
diff --git a/alignment/CMakeLists.txt b/alignment/CMakeLists.txt
diff --git a/alignment/GraphAlignmentOperations.cpp b/alignment/GraphAlignmentOperations.cpp
diff --git a/alignment/GraphAlignmentOperations.hh b/alignment/GraphAlignmentOperations.hh
diff --git a/alignment/GreedyAlignmentIntersector.cpp b/alignment/GreedyAlignmentIntersector.cpp
diff --git a/alignment/GreedyAlignmentIntersector.hh b/alignment/GreedyAlignmentIntersector.hh
diff --git a/alignment/HighQualityBaseRunFinder.cpp b/alignment/HighQualityBaseRunFinder.cpp
diff --git a/alignment/HighQualityBaseRunFinder.hh b/alignment/HighQualityBaseRunFinder.hh
diff --git a/alignment/SoftclippingAligner.cpp b/alignment/SoftclippingAligner.cpp
diff --git a/alignment/SoftclippingAligner.hh b/alignment/SoftclippingAligner.hh
diff --git a/alignment/tests/AlignmentTweakersTest.cpp b/alignment/tests/AlignmentTweakersTest.cpp
diff --git a/alignment/tests/CMakeLists.txt b/alignment/tests/CMakeLists.txt
diff --git a/alignment/tests/GraphAlignmentOperationsTest.cpp b/alignment/tests/GraphAlignmentOperationsTest.cpp
diff --git a/alignment/tests/GreedyAlignmentIntersectorTest.cpp b/alignment/tests/GreedyAlignmentIntersectorTest.cpp
diff --git a/alignment/tests/HighQualityBaseRunFinderTest.cpp b/alignment/tests/HighQualityBaseRunFinderTest.cpp
diff --git a/alignment/tests/SoftclippingAlignerTest.cpp b/alignment/tests/SoftclippingAlignerTest.cpp
diff --git a/classification/AlignmentClassifier.cpp b/classification/AlignmentClassifier.cpp
diff --git a/classification/AlignmentClassifier.hh b/classification/AlignmentClassifier.hh
diff --git a/classification/CMakeLists.txt b/classification/CMakeLists.txt
diff --git a/classification/ClassifierOfAlignmentsToVariant.cpp b/classification/ClassifierOfAlignmentsToVariant.cpp
diff --git a/classification/ClassifierOfAlignmentsToVariant.hh b/classification/ClassifierOfAlignmentsToVariant.hh
diff --git a/classification/tests/AlignmentClassifierTest.cpp b/classification/tests/AlignmentClassifierTest.cpp
diff --git a/classification/tests/AlignmentSummaryTest.cpp b/classification/tests/AlignmentSummaryTest.cpp
@@ -28,7 +28,6 @@
 
 using namespace ehunter;
 
-using reads::Read;
 using std::map;
 using std::vector;
 

diff --git a/classification/tests/CMakeLists.txt b/classification/tests/CMakeLists.txt
diff --git a/classification/tests/ClassifierOfAlignmentsToVariantTest.cpp b/classification/tests/ClassifierOfAlignmentsToVariantTest.cpp
diff --git a/cmake/google_test.cmake b/cmake/google_test.cmake
diff --git a/common/CMakeLists.txt b/common/CMakeLists.txt
diff --git a/common/Common.cpp b/common/Common.cpp
diff --git a/common/Common.hh b/common/Common.hh
diff --git a/common/CountTable.cpp b/common/CountTable.cpp
diff --git a/common/CountTable.hh b/common/CountTable.hh
diff --git a/common/GenomicRegion.cpp b/common/GenomicRegion.cpp
@@ -22,55 +22,34 @@
 
 #include "common/GenomicRegion.hh"
 
-#include <boost/algorithm/string/classification.hpp>
-#include <boost/algorithm/string/split.hpp>
-#include <boost/lexical_cast.hpp>
-
 #include <limits>
 #include <sstream>
 #include <stdexcept>
-#include <string>
-#include <vector>
+
+#include <boost/algorithm/string/classification.hpp>
+#include <boost/algorithm/string/split.hpp>
 
 using std::istream;
 using std::ostream;
 using std::string;
+using std::unordered_map;
 using std::vector;
 
-using boost::lexical_cast;
-using boost::algorithm::is_any_of;
-using boost::algorithm::split;
-
 namespace ehunter
 {
 
-Region::Region(const std::string chrom, int64_t start, int64_t end)
-    : chrom_(chrom)
+GenomicRegion::GenomicRegion(int32_t contigIndex, int64_t start, int64_t end)
+    : contigIndex_(contigIndex)
     , start_(start)
     , end_(end)
 {
 }
 
-Region::Region(const std::string encoding)
-{
-    vector<string> components;
-    boost::algorithm::split(components, encoding, is_any_of(":-"));
-
-    if (components.size() != 3)
-    {
-        throw std::logic_error("Unexpected range format: " + encoding);
-    }
-
-    chrom_ = components[0];
-    start_ = lexical_cast<int64_t>(components[1]);
-    end_ = lexical_cast<int64_t>(components[2]);
-}
-
-bool Region::operator<(const Region& other) const
+bool GenomicRegion::operator<(const GenomicRegion& other) const
 {
-    if (chrom_ != other.chrom_)
+    if (contigIndex_ != other.contigIndex_)
     {
-        return chrom_ < other.chrom_;
+        return contigIndex_ < other.contigIndex_;
     }
 
     if (start_ != other.start_)
@@ -81,9 +60,9 @@ bool Region::operator<(const Region& other) const
     return end_ < other.end_;
 }
 
-bool Region::Overlaps(const Region& other) const
+bool GenomicRegion::overlaps(const GenomicRegion& other) const
 {
-    if (chrom_ != other.chrom_)
+    if (contigIndex_ != other.contigIndex_)
     {
         return false;
     }
@@ -94,9 +73,9 @@ bool Region::Overlaps(const Region& other) const
     return leftBound <= rightBound;
 }
 
-int64_t Region::Distance(const Region& other) const
+int64_t GenomicRegion::distance(const GenomicRegion& other) const
 {
-    if (chrom_ != other.chrom_)
+    if (contigIndex_ != other.contigIndex_)
     {
         return std::numeric_limits<int64_t>::max();
     }
@@ -114,7 +93,7 @@ int64_t Region::Distance(const Region& other) const
     return 0;
 }
 
-vector<Region> merge(vector<Region> regions, int maxMergeDist)
+vector<GenomicRegion> merge(vector<GenomicRegion> regions, int maxMergeDist)
 {
     if (regions.empty())
     {
@@ -123,12 +102,12 @@ vector<Region> merge(vector<Region> regions, int maxMergeDist)
 
     std::sort(regions.begin(), regions.end());
 
-    Region mergedRegion = regions.front();
-    vector<Region> mergedRegions;
+    GenomicRegion mergedRegion = regions.front();
+    vector<GenomicRegion> mergedRegions;
 
     for (const auto& currentRegion : regions)
     {
-        if (currentRegion.Distance(mergedRegion) <= maxMergeDist)
+        if (currentRegion.distance(mergedRegion) <= maxMergeDist)
         {
             const int64_t furthestEnd = std::max<int64_t>(mergedRegion.end(), currentRegion.end());
             mergedRegion.setEnd(furthestEnd);
@@ -148,27 +127,45 @@ vector<Region> merge(vector<Region> regions, int maxMergeDist)
     return mergedRegions;
 }
 
-const string Region::ToString() const
-{
-    std::ostringstream out;
-    out << *this;
-    return out.str();
-}
-
 // Returns the range extended by flankSize upstream and downstream.
 // NOTE: The right boundary of the extended region may stick past chromosome
 // end.
-Region Region::extend(int length) const
+GenomicRegion GenomicRegion::extend(int length) const
 {
-    const int64_t new_start = start_ >= length ? (start_ - length) : 0;
-    const int64_t new_end = end_ + length;
-    return Region(chrom_, new_start, new_end);
+    const int64_t newStart = start_ >= length ? (start_ - length) : 0;
+    const int64_t newEnd = end_ + length;
+    return GenomicRegion(contigIndex_, newStart, newEnd);
 }
 
-std::ostream& operator<<(std::ostream& out, const Region& region)
+std::ostream& operator<<(std::ostream& out, const GenomicRegion& region)
 {
-    out << region.chrom_ << ":" << region.start_ << "-" << region.end_;
+    out << "(" << region.contigIndex_ << "):" << region.start_ << "-" << region.end_;
     return out;
 }
 
+string encode(const ReferenceContigInfo& contigInfo, const GenomicRegion& region)
+{
+    const auto& contigName = contigInfo.getContigName(region.contigIndex());
+    return contigName + ":" + std::to_string(region.start()) + "-" + std::to_string(region.end());
+}
+
+GenomicRegion decode(const ReferenceContigInfo& contigInfo, const string& encoding)
+{
+    vector<string> components;
+    boost::algorithm::split(components, encoding, boost::algorithm::is_any_of(":-"));
+
+    if (components.size() != 3)
+    {
+        throw std::logic_error("Unexpected range format: " + encoding);
+    }
+
+    const auto& contigName = components[0];
+    int32_t contigIndex = contigInfo.getContigId(contigName);
+
+    int64_t start = std::stoi(components[1]);
+    int64_t end = std::stoi(components[2]);
+
+    return GenomicRegion(contigIndex, start, end);
+}
+
 }
diff --git a/common/GenomicRegion.hh b/common/GenomicRegion.hh
@@ -24,48 +24,57 @@
 
 #include <iostream>
 #include <string>
+#include <unordered_map>
 #include <vector>
 
+#include "common/ReferenceContigInfo.hh"
+
 namespace ehunter
 {
 
-class Region
+// Represents a contiguous region of a genome using 0-based half-open coordinates
+class GenomicRegion
 {
 public:
-    friend std::ostream& operator<<(std::ostream& out, const Region& region);
+    friend std::ostream& operator<<(std::ostream& out, const GenomicRegion& region);
 
-    Region(const std::string chrom, int64_t start, int64_t end);
-    explicit Region(const std::string encoding);
+    GenomicRegion(const int32_t contigIndex, int64_t start, int64_t end);
 
-    bool operator<(const Region& other) const;
+    bool operator<(const GenomicRegion& other) const;
 
-    bool Overlaps(const Region& other) const;
-    int64_t Distance(const Region& other) const;
+    bool overlaps(const GenomicRegion& other) const;
+    int64_t distance(const GenomicRegion& other) const;
 
-    const std::string& chrom() const { return chrom_; }
+    int32_t contigIndex() const { return contigIndex_; }
     int64_t start() const { return start_; }
     int64_t end() const { return end_; }
-    int64_t length() const { return end_ - start_ + 1; }
+    int64_t length() const { return end_ - start_; }
 
-    void setChrom(const std::string& chrom) { chrom_ = chrom; }
+    void setContigId(int32_t contigIndex) { contigIndex_ = contigIndex; }
     void setStart(int64_t start) { start_ = start; }
     void setEnd(int64_t end) { end_ = end; }
-    bool operator==(const Region& other) const
+
+    bool operator==(const GenomicRegion& other) const
     {
-        return chrom_ == other.chrom_ && start_ == other.start_ && end_ == other.end_;
+        return contigIndex_ == other.contigIndex_ && start_ == other.start_ && end_ == other.end_;
     }
-    bool operator!=(const Region& other) const { return !(*this == other); }
 
-    Region extend(int length) const;
-    const std::string ToString() const;
+    bool operator!=(const GenomicRegion& other) const { return !(*this == other); }
+
+    GenomicRegion extend(int length) const;
 
 private:
-    std::string chrom_;
+    int32_t contigIndex_;
     int64_t start_;
     int64_t end_;
 };
 
-std::vector<Region> merge(std::vector<Region> regions, int maxMergeDist = 500);
-std::ostream& operator<<(std::ostream& out, const Region& region);
+using GenomicRegionCatalog = std::unordered_map<std::string, GenomicRegion>;
+
+std::ostream& operator<<(std::ostream& out, const GenomicRegion& region);
+std::vector<GenomicRegion> merge(std::vector<GenomicRegion> regions, int maxMergeDist = 500);
+
+std::string encode(const ReferenceContigInfo& contigInfo, const GenomicRegion& region);
+GenomicRegion decode(const ReferenceContigInfo& contigInfo, const std::string& encoding);
 
 }