Tools: Function for converting a nucleotide to an integer.

Use bitwise operations which avoids branching with if/switch.
Tests added

algo/core/tools.cpp

@@ -119,6 +119,18 @@ string complement(const string &dna) {
   return comp;
 }
 
+int nuc_to_int(char nuc) {
+  // A : 1000 0001
+  // C : 1000 0011
+  // G : 1000 0111
+  // T : 1001 1000
+  // pos :    3210
+  // Bit de poids fort : b_3 ou b_2
+  // Bit de poids faible : xor entre b_3, b_2 et b_1
+  return (((nuc & 4) >> 1) | ((nuc & 8) >> 2)) // poids fort
+    | (((nuc & 8) >> 3) ^ ((nuc & 4) >> 2) ^ ((nuc & 2) >> 1));
+}
+
 Sequence create_sequence(string label_full, string label, string sequence, string quality) {
   Sequence seq;
   seq.label_full = label_full;

algo/core/tools.h

@@ -58,6 +58,11 @@ char complement_nucleotide(char nuc);
  */
 string complement(const string &dna);
 
+/**
+ * @pre nuc est un nucléotide en majuscule (A, C, G ou T)
+ * @return le code entier du nucléotide (respectivement 0, 1, 2 ou 3)
+ */
+int nuc_to_int(char nuc);
 
 string extract_from_label(string str, int field, string separator);
 

algo/tests/testTools.cpp

@@ -38,8 +38,16 @@ void testCreateSequence() {
   TAP_TEST(seq2.quality == "", TEST_CREATE_SEQUENCE_QUALITY, "");
 }
 
+void testNucToInt() {
+  TAP_TEST(nuc_to_int('A') == 0, TEST_NUC_TO_INT, "");
+  TAP_TEST(nuc_to_int('C') == 1, TEST_NUC_TO_INT, "");
+  TAP_TEST(nuc_to_int('G') == 2, TEST_NUC_TO_INT, "");
+  TAP_TEST(nuc_to_int('T') == 3, TEST_NUC_TO_INT, "");
+}
+
 void testTools() {
   testFasta1();
   testRevcomp();
   testCreateSequence();
+  testNucToInt();
 }

algo/tests/tests.h

@@ -12,6 +12,7 @@ enum {
   TEST_CREATE_SEQUENCE_LABEL,
   TEST_CREATE_SEQUENCE_SEQUENCE,
   TEST_CREATE_SEQUENCE_QUALITY,
+  TEST_NUC_TO_INT,
 
   /* Storage tests */
   TEST_ARRAY_KMERSTORE,
@@ -97,6 +98,7 @@ inline void declare_tests() {
   RECORD_TAP_TEST(TEST_CREATE_SEQUENCE_LABEL, "create_sequence: label field");
   RECORD_TAP_TEST(TEST_CREATE_SEQUENCE_SEQUENCE, "create_sequence: sequence field");
   RECORD_TAP_TEST(TEST_CREATE_SEQUENCE_QUALITY, "create_sequence: quality field");
+  RECORD_TAP_TEST(TEST_NUC_TO_INT, "nuc_to_int()");
 
   RECORD_TAP_TEST(TEST_ARRAY_KMERSTORE, "Testing ArrayKmerStore");
   RECORD_TAP_TEST(TEST_MAP_KMERSTORE, "Testing MapKmerStore");