[TEST] Add seqan3::simd_dna4

Co-authored-by: Enrico Seiler <enrico.seiler@hotmail.de>

test/include/seqan3/test/performance/simd_dna4.hpp

@@ -0,0 +1,102 @@
+// -----------------------------------------------------------------------------------------------------
+// Copyright (c) 2006-2022, Knut Reinert & Freie Universität Berlin
+// Copyright (c) 2016-2022, Knut Reinert & MPI für molekulare Genetik
+// This file may be used, modified and/or redistributed under the terms of the 3-clause BSD-License
+// shipped with this file and also available at: https://github.com/seqan/seqan3/blob/master/LICENSE.md
+// -----------------------------------------------------------------------------------------------------
+
+#pragma once
+
+#include <seqan3/alphabet/nucleotide/nucleotide_base.hpp>
+#include <seqan3/alphabet/nucleotide/rna4.hpp>
+
+// Conditionally add the seqan3 namespace:
+// unit/performance tests: in seqan3 namespace
+// cookbook entry: no namespace
+#ifndef SEQAN3_USE_NAMESPACE
+#    define SEQAN3_USE_NAMESPACE 1
+#endif
+
+#if SEQAN3_USE_NAMESPACE
+namespace seqan3
+{
+#endif
+/* See discussion here: https://github.com/seqan/seqan3/issues/1970
+ *
+ * If AVX2 is available, this significantly improves the performance
+ * (roughly 5 times faster). But if no AVX2 is available. this decreases the
+ * performance (roughly 3 times slower).
+ */
+//![cookbook]
+class simd_dna4 : public seqan3::nucleotide_base<simd_dna4, 256>
+{
+private:
+    using base_t = seqan3::nucleotide_base<simd_dna4, 256>;
+
+    friend base_t;         // nucleotide_base
+    friend base_t::base_t; // alphabet_base
+    friend seqan3::rna4;
+
+public:
+    constexpr simd_dna4() noexcept = default;
+    constexpr simd_dna4(simd_dna4 const &) noexcept = default;
+    constexpr simd_dna4(simd_dna4 &&) noexcept = default;
+    constexpr simd_dna4 & operator=(simd_dna4 const &) noexcept = default;
+    constexpr simd_dna4 & operator=(simd_dna4 &&) noexcept = default;
+    ~simd_dna4() noexcept = default;
+
+    template <std::same_as<seqan3::rna4> t> // template parameter t to accept incomplete type
+    constexpr simd_dna4(t const r) noexcept
+    {
+        assign_rank(r.to_rank());
+    }
+
+    using base_t::assign_rank;
+    using base_t::base_t;
+    using base_t::to_rank;
+
+    static constexpr uint8_t alphabet_size = 4;
+
+    constexpr simd_dna4 & assign_char(char_type const c) noexcept
+    {
+        char_type const upper_case_char = c & 0b0101'1111;
+        rank_type rank = (upper_case_char == 'T') * 3 + (upper_case_char == 'G') * 2 + (upper_case_char == 'C');
+        return assign_rank(rank);
+    }
+
+    constexpr char_type to_char() const noexcept
+    {
+        rank_type const rank = to_rank();
+        switch (rank)
+        {
+        case 0u:
+            return 'A';
+        case 1u:
+            return 'C';
+        case 2u:
+            return 'G';
+        default:
+            return 'T';
+        }
+    }
+
+    constexpr simd_dna4 complement() const noexcept
+    {
+        rank_type rank{to_rank()};
+        rank ^= 0b11;
+        simd_dna4 ret{};
+        return ret.assign_rank(rank);
+    }
+
+    static constexpr bool char_is_valid(char_type const c) noexcept
+    {
+        char_type const upper_case_char = c & 0b0101'1111;
+        return (upper_case_char == 'A') || (upper_case_char == 'C') || (upper_case_char == 'G')
+            || (upper_case_char == 'T');
+    }
+};
+//![cookbook]
+
+#if SEQAN3_USE_NAMESPACE
+}
+#endif

test/performance/alphabet/alphabet_assign_char_benchmark.cpp

@@ -12,6 +12,7 @@
 #include <numeric>
 
 #include <seqan3/alphabet/all.hpp>
+#include <seqan3/test/performance/simd_dna4.hpp>
 #include <seqan3/test/seqan2.hpp>
 
 #if SEQAN3_HAS_SEQAN2
@@ -38,6 +39,7 @@ void assign_char(benchmark::State & state)
 BENCHMARK_TEMPLATE(assign_char, seqan3::gap);
 BENCHMARK_TEMPLATE(assign_char, seqan3::dna4);
 BENCHMARK_TEMPLATE(assign_char, seqan3::rna4);
+BENCHMARK_TEMPLATE(assign_char, seqan3::simd_dna4);
 BENCHMARK_TEMPLATE(assign_char, seqan3::dna5);
 BENCHMARK_TEMPLATE(assign_char, seqan3::rna5);
 BENCHMARK_TEMPLATE(assign_char, seqan3::dna15);

test/performance/alphabet/alphabet_assign_rank_benchmark.cpp

@@ -11,6 +11,7 @@
 #include <cstring>
 
 #include <seqan3/alphabet/all.hpp>
+#include <seqan3/test/performance/simd_dna4.hpp>
 #include <seqan3/test/seqan2.hpp>
 
 #if SEQAN3_HAS_SEQAN2
@@ -45,6 +46,7 @@ void assign_rank(benchmark::State & state)
 BENCHMARK_TEMPLATE(assign_rank, seqan3::gap);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::dna4);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::rna4);
+BENCHMARK_TEMPLATE(assign_rank, seqan3::simd_dna4);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::dna5);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::rna5);
 BENCHMARK_TEMPLATE(assign_rank, seqan3::dna15);

test/performance/alphabet/alphabet_to_char_benchmark.cpp

@@ -11,6 +11,7 @@
 #include <cstring>
 
 #include <seqan3/alphabet/all.hpp>
+#include <seqan3/test/performance/simd_dna4.hpp>
 #include <seqan3/test/seqan2.hpp>
 
 #if SEQAN3_HAS_SEQAN2
@@ -58,6 +59,7 @@ void to_char(benchmark::State & state)
 BENCHMARK_TEMPLATE(to_char, seqan3::gap);
 BENCHMARK_TEMPLATE(to_char, seqan3::dna4);
 BENCHMARK_TEMPLATE(to_char, seqan3::rna4);
+BENCHMARK_TEMPLATE(to_char, seqan3::simd_dna4);
 BENCHMARK_TEMPLATE(to_char, seqan3::dna5);
 BENCHMARK_TEMPLATE(to_char, seqan3::rna5);
 BENCHMARK_TEMPLATE(to_char, seqan3::dna15);

test/performance/alphabet/alphabet_to_rank_benchmark.cpp

@@ -10,6 +10,7 @@
 #include <algorithm>
 
 #include <seqan3/alphabet/all.hpp>
+#include <seqan3/test/performance/simd_dna4.hpp>
 #include <seqan3/test/seqan2.hpp>
 
 #if SEQAN3_HAS_SEQAN2
@@ -52,6 +53,7 @@ void to_rank(benchmark::State & state)
 BENCHMARK_TEMPLATE(to_rank, seqan3::gap);
 BENCHMARK_TEMPLATE(to_rank, seqan3::dna4);
 BENCHMARK_TEMPLATE(to_rank, seqan3::rna4);
+BENCHMARK_TEMPLATE(to_rank, seqan3::simd_dna4);
 BENCHMARK_TEMPLATE(to_rank, seqan3::dna5);
 BENCHMARK_TEMPLATE(to_rank, seqan3::rna5);
 BENCHMARK_TEMPLATE(to_rank, seqan3::dna15);

test/unit/alphabet/nucleotide/CMakeLists.txt

@@ -7,3 +7,4 @@ seqan3_test (rna4_test.cpp)
 seqan3_test (rna5_test.cpp)
 seqan3_test (rna15_test.cpp)
 seqan3_test (nucleotide_conversion_integration_test.cpp)
+seqan3_test (simd_dna4_test.cpp)

test/unit/alphabet/nucleotide/simd_dna4_test.cpp

@@ -0,0 +1,35 @@
+// -----------------------------------------------------------------------------------------------------
+// Copyright (c) 2006-2022, Knut Reinert & Freie Universität Berlin
+// Copyright (c) 2016-2022, Knut Reinert & MPI für molekulare Genetik
+// This file may be used, modified and/or redistributed under the terms of the 3-clause BSD-License
+// shipped with this file and also available at: https://github.com/seqan/seqan3/blob/master/LICENSE.md
+// -----------------------------------------------------------------------------------------------------
+
+#include <seqan3/core/debug_stream/range.hpp>
+#include <seqan3/test/performance/simd_dna4.hpp>
+#include <seqan3/utility/char_operations/predicate.hpp>
+
+#include "../alphabet_constexpr_test_template.hpp"
+#include "../alphabet_test_template.hpp"
+#include "../semi_alphabet_constexpr_test_template.hpp"
+#include "../semi_alphabet_test_template.hpp"
+#include "nucleotide_test_template.hpp"
+
+INSTANTIATE_TYPED_TEST_SUITE_P(simdifyable_dna4, alphabet, seqan3::simd_dna4, );
+INSTANTIATE_TYPED_TEST_SUITE_P(simdifyable_dna4, semi_alphabet_test, seqan3::simd_dna4, );
+INSTANTIATE_TYPED_TEST_SUITE_P(simdifyable_dna4, alphabet_constexpr, seqan3::simd_dna4, );
+INSTANTIATE_TYPED_TEST_SUITE_P(simdifyable_dna4, semi_alphabet_constexpr, seqan3::simd_dna4, );
+INSTANTIATE_TYPED_TEST_SUITE_P(simdifyable_dna4, nucleotide, seqan3::simd_dna4, );
+
+TEST(simdifyable_dna4, to_char_assign_char)
+{
+    EXPECT_EQ(seqan3::to_char(seqan3::simd_dna4{}.assign_char('A')), 'A');
+    EXPECT_EQ(seqan3::to_char(seqan3::simd_dna4{}.assign_char('C')), 'C');
+    EXPECT_EQ(seqan3::to_char(seqan3::simd_dna4{}.assign_char('G')), 'G');
+    EXPECT_EQ(seqan3::to_char(seqan3::simd_dna4{}.assign_char('T')), 'T');
+
+    EXPECT_EQ(seqan3::to_char(seqan3::simd_dna4{}.assign_char('a')), 'A');
+    EXPECT_EQ(seqan3::to_char(seqan3::simd_dna4{}.assign_char('c')), 'C');
+    EXPECT_EQ(seqan3::to_char(seqan3::simd_dna4{}.assign_char('g')), 'G');
+    EXPECT_EQ(seqan3::to_char(seqan3::simd_dna4{}.assign_char('t')), 'T');
+}