Tools: revcomp_int performs revcomp on an integer.

Revcomp directly on an integer.
Tests added

algo/core/tools.cpp

@@ -207,6 +207,17 @@ string revcomp(const string &dna, bool do_revcomp) {
   return rcomp;
 }
 
+int revcomp_int(int word, int size) {
+  int revcomp = 0;
+  while (size) {
+    revcomp <<= 2;
+    revcomp |= (word & 3) ^ 3;
+    word >>= 2;
+    size--;
+  }
+  return revcomp;
+}
+
 string reverse(const string &text) {
   return string(text.rbegin(), text.rend());
 }

algo/core/tools.h

@@ -84,6 +84,12 @@ int remove_trailing_whitespaces(string &str);
  */
 string revcomp(const string &dna, bool do_revcomp = true);
 
+/**
+ * @return the int value corresponding to the revcomp of the DNA sequence
+ *         represented by word, whose length (in number of nucleotides) is size.
+ */
+int revcomp_int(int word, int size);
+
 /**
  * @return the reverse of text (ie. text read from right to left)
  */

algo/tests/testTools.cpp

@@ -52,10 +52,48 @@ void testDNAToInt() {
   TAP_TEST(dna_to_int("TTTT", 4) == 255, TEST_DNA_TO_INT, "");
 }
 
+void testRevcompInt() {
+  TAP_TEST(revcomp_int(dna_to_int("AA", 2), 2) == dna_to_int("TT", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("AC", 2), 2) == dna_to_int("GT", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("AG", 2), 2) == dna_to_int("CT", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("AT", 2), 2) == dna_to_int("AT", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("CA", 2), 2) == dna_to_int("TG", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("CC", 2), 2) == dna_to_int("GG", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("CG", 2), 2) == dna_to_int("CG", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("CT", 2), 2) == dna_to_int("AG", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("GA", 2), 2) == dna_to_int("TC", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("GC", 2), 2) == dna_to_int("GC", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("GT", 2), 2) == dna_to_int("AC", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("TA", 2), 2) == dna_to_int("TA", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("TC", 2), 2) == dna_to_int("GA", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("TG", 2), 2) == dna_to_int("CA", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("TT", 2), 2) == dna_to_int("AA", 2),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("AAAAAAA", 7), 7) == dna_to_int("TTTTTTT", 7),
+           TEST_REVCOMP_INT, "");
+  TAP_TEST(revcomp_int(dna_to_int("ATTAGGA", 7), 7) == dna_to_int("TCCTAAT", 7),
+           TEST_REVCOMP_INT, "revcomp: " << revcomp_int(dna_to_int("ATTAGGA", 7), 7) <<", dna_to_int: " << dna_to_int("TCCTAAT", 7));
+}
+
 void testTools() {
   testFasta1();
   testRevcomp();
   testCreateSequence();
   testNucToInt();
   testDNAToInt();
+  testRevcompInt();
 }

algo/tests/tests.h

@@ -14,6 +14,7 @@ enum {
   TEST_CREATE_SEQUENCE_QUALITY,
   TEST_NUC_TO_INT,
   TEST_DNA_TO_INT,
+  TEST_REVCOMP_INT,
 
   /* Storage tests */
   TEST_ARRAY_KMERSTORE,
@@ -101,6 +102,7 @@ inline void declare_tests() {
   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_DNA_TO_INT, "dna_to_int()");
+  RECORD_TAP_TEST(TEST_REVCOMP_INT, "revcomp_int()");
 
   RECORD_TAP_TEST(TEST_ARRAY_KMERSTORE, "Testing ArrayKmerStore");
   RECORD_TAP_TEST(TEST_MAP_KMERSTORE, "Testing MapKmerStore");