Add Kmer Color Gamut serialization

Author: lczech

lib/genesis/sequence/kmer/color_gamut.hpp

@@ -213,7 +213,7 @@ class KmerColorGamut
         return lookup_;
     }
 
-    utils::Matrix<size_t> const& get_gamut() const
+    utils::Matrix<size_t> const& get_gamut_matrix() const
     {
         return gamut_;
     }

lib/genesis/sequence/kmer/color_gamut_functions.cpp

@@ -33,6 +33,7 @@
 #include "genesis/taxonomy/functions/taxonomy.hpp"
 #include "genesis/taxonomy/taxon.hpp"
 #include "genesis/taxonomy/taxonomy.hpp"
+#include "genesis/utils/containers/matrix/operators.hpp"
 #include "genesis/utils/core/algorithm.hpp"
 #include "genesis/utils/io/deserializer.hpp"
 #include "genesis/utils/io/serializer.hpp"
@@ -482,7 +483,7 @@ std::string print_kmer_color_lookup( KmerColorGamut const& gamut )
 
 std::string print_kmer_color_gamut( KmerColorGamut const& gamut )
 {
-    auto const gamut_matrix = gamut.get_gamut();
+    auto const gamut_matrix = gamut.get_gamut_matrix();
     if( gamut_matrix.empty() ) {
         return "";
     }
@@ -517,7 +518,7 @@ std::string print_kmer_color_gamut_summary( KmerColorGamut const& gamut )
     assert( gamut.get_color_list().size() == gamut.get_color_lookup().size() );
 
     // Report how many of the gamut are real and how many are imaginar numbers
-    auto const& gamut_matrix = gamut.get_gamut();
+    auto const& gamut_matrix = gamut.get_gamut_matrix();
     auto const& stats = gamut.get_gamut_statistics();
     auto const gamut_size = gamut_matrix.rows() * gamut_matrix.cols();
     auto const gamut_empty = gamut_size - ( stats.real_color_count + stats.imag_color_count);
@@ -558,6 +559,56 @@ std::string print_kmer_color_gamut_summary( KmerColorGamut const& gamut )
     return ss.str();
 }
 
+// =================================================================================================
+//     Serialization
+// =================================================================================================
+
+void serialize_kmer_color_gamut_colors(
+    KmerColorGamut const& gamut,
+    std::shared_ptr<utils::BaseOutputTarget> output_target
+) {
+    // We need to iterate through the color vector manually here,
+    // as the Bitvectors are wrapped in a struct for future compatibility.
+    auto ser = utils::Serializer( output_target );
+    auto const& color_list = gamut.get_color_list();
+    ser << color_list.size();
+    for( auto const& color : color_list ) {
+        ser << color.elements;
+    }
+}
+
+void serialize_kmer_color_gamut_matrix(
+    KmerColorGamut const& gamut,
+    std::shared_ptr<utils::BaseOutputTarget> output_target
+) {
+    // Functionality for std::vector and Matrix is already
+    // implemented in the respective classes and functions.
+    auto ser = utils::Serializer( output_target );
+    ser << gamut.get_gamut_matrix();
+}
+
+std::vector<utils::Bitvector> deserialize_kmer_color_gamut_colors(
+    std::shared_ptr<utils::BaseInputSource> input_source
+) {
+    // Functionality for std::vector and Bitvector is already
+    // implemented in the respective classes and functions.
+    std::vector<utils::Bitvector> result;
+    auto deser = utils::Deserializer( input_source );
+    deser >> result;
+    return result;
+}
+
+utils::Matrix<size_t> deserialize_kmer_color_gamut_matrix(
+    std::shared_ptr<utils::BaseInputSource> input_source
+) {
+    // Functionality for std::vector and Matrix is already
+    // implemented in the respective classes and functions.
+    utils::Matrix<size_t> result;
+    auto deser = utils::Deserializer( input_source );
+    deser >> result;
+    return result;
+}
+
 } // namespace sequence
 } // namespace genesis
 

lib/genesis/sequence/kmer/color_gamut_functions.hpp

@@ -33,7 +33,10 @@
 
 #include "genesis/sequence/kmer/color_gamut.hpp"
 #include "genesis/taxonomy/taxonomy.hpp"
+#include "genesis/utils/containers/matrix.hpp"
 #include "genesis/utils/io/input_source.hpp"
+#include "genesis/utils/io/input_source.hpp"
+#include "genesis/utils/io/output_target.hpp"
 #include "genesis/utils/io/output_target.hpp"
 #include "genesis/utils/math/bitvector.hpp"
 #include "genesis/utils/math/bitvector/functions.hpp"
@@ -138,6 +141,24 @@ std::string print_kmer_color_lookup( KmerColorGamut const& gamut );
 std::string print_kmer_color_gamut( KmerColorGamut const& gamut );
 std::string print_kmer_color_gamut_summary( KmerColorGamut const& gamut );
 
+void serialize_kmer_color_gamut_colors(
+    KmerColorGamut const& gamut,
+    std::shared_ptr<utils::BaseOutputTarget> output_target
+);
+
+void serialize_kmer_color_gamut_matrix(
+    KmerColorGamut const& gamut,
+    std::shared_ptr<utils::BaseOutputTarget> output_target
+);
+
+std::vector<utils::Bitvector> deserialize_kmer_color_gamut_colors(
+    std::shared_ptr<utils::BaseInputSource> input_source
+);
+
+utils::Matrix<size_t> deserialize_kmer_color_gamut_matrix(
+    std::shared_ptr<utils::BaseInputSource> input_source
+);
+
 } // namespace sequence
 } // namespace genesis
 

test/src/sequence/kmer_color_gamut.cpp

@@ -51,6 +51,7 @@
 
 #include <algorithm>
 #include <atomic>
+#include <cstdio>
 #include <future>
 #include <iomanip>
 #include <ios>
@@ -127,6 +128,7 @@ TEST( KmerColorGamut, Basics )
 
 TEST( KmerColorGamut, Example )
 {
+    NEEDS_TEST_DATA;
     auto const p = 10;
     auto const r = 60;
 
@@ -252,8 +254,8 @@ TEST( KmerColorGamut, Example )
 
     // We added 12 imaginary colors now, and have switched to gamut now.
     EXPECT_EQ( initial_gamut_size + 5 * 8, gamut.get_color_list().size() );
-    EXPECT_EQ( r, gamut.get_gamut().rows() );
-    EXPECT_EQ( p, gamut.get_gamut().cols() );
+    EXPECT_EQ( r, gamut.get_gamut_matrix().rows() );
+    EXPECT_EQ( p, gamut.get_gamut_matrix().cols() );
     EXPECT_EQ(  0, gamut.get_gamut_statistics().real_color_count );
     EXPECT_EQ( 12, gamut.get_gamut_statistics().imag_color_count );
 
@@ -263,6 +265,25 @@ TEST( KmerColorGamut, Example )
     // LOG_DBG << print_kmer_color_lookup( gamut );
     // LOG_DBG << print_kmer_color_gamut( gamut );
     // LOG_DBG << print_kmer_color_gamut_summary( gamut );
+
+    // Test the serialization
+    auto const colors_file = environment->data_dir + "sequence/gamut_example_colors.bin";
+    auto const matrix_file = environment->data_dir + "sequence/gamut_example_matrix.bin";
+    serialize_kmer_color_gamut_colors( gamut, to_file( colors_file ));
+    serialize_kmer_color_gamut_matrix( gamut, to_file( matrix_file ));
+    auto const deser_colors = deserialize_kmer_color_gamut_colors( from_file( colors_file ));
+    auto const deser_matrix = deserialize_kmer_color_gamut_matrix( from_file( matrix_file ));
+    EXPECT_EQ( 0, std::remove( colors_file.c_str() ));
+    EXPECT_EQ( 0, std::remove( matrix_file.c_str() ));
+    EXPECT_EQ( initial_gamut_size + 5 * 8, deser_colors.size() );
+    EXPECT_EQ( gamut.get_color_list().size(), deser_colors.size() );
+    for( size_t i = 0; i < deser_colors.size(); ++i ) {
+        EXPECT_EQ( p, deser_colors[i].size() );
+        EXPECT_EQ( gamut.get_color_at(i).elements, deser_colors[i] );
+    }
+    EXPECT_EQ( r, deser_matrix.rows() );
+    EXPECT_EQ( p, deser_matrix.cols() );
+    EXPECT_EQ( gamut.get_gamut_matrix(), deser_matrix );
 }
 
 // -------------------------------------------------------------------------
@@ -319,10 +340,10 @@ TEST( KmerColorGamut, Random )
     // We have run the loop enough that we can be virtually sure to have requested
     // every color in the 8 bit range that we are using here, so the gamut should be full.
     EXPECT_EQ( 16, gamut.get_color_list().size() );
-    EXPECT_EQ( 16, gamut.get_gamut().rows() );
-    EXPECT_EQ(  8, gamut.get_gamut().cols() );
+    EXPECT_EQ( 16, gamut.get_gamut_matrix().rows() );
+    EXPECT_EQ(  8, gamut.get_gamut_matrix().cols() );
     auto img_idx_counts = std::vector<size_t>( 16, 0 );
-    for( auto c : gamut.get_gamut() ) {
+    for( auto c : gamut.get_gamut_matrix() ) {
         ++img_idx_counts[c];
     }