algo/core/tools.{h,cpp} + tests: Give trimSequence a region that should not be trimmed

algo/core/tools.cpp

@@ -372,7 +372,8 @@ double nChoosek(unsigned n, unsigned k)
     return nChoosek_stored[n][k];
 }
 
-void trimSequence(string &sequence, size_t &start_pos, size_t &length) {
+void trimSequence(string &sequence, size_t &start_pos, size_t &length,
+                  size_t required_start, size_t required_length) {
   float prefix_score = 0;
   float suffix_score = 0;
   size_t start_bad_suffix = 0;
@@ -425,8 +426,8 @@ void trimSequence(string &sequence, size_t &start_pos, size_t &length) {
     }
   }
 
-  start_pos = max_start_factor;
-  length = max_factor_length;
+  start_pos = min(required_start, max_start_factor);
+  length = max(required_length, max_factor_length);
 }
 
 

algo/core/tools.h

@@ -259,9 +259,13 @@ double nChoosek(unsigned n, unsigned k);
  * More precisely, the purpose of the function is to find the longest
  * substring whose prefixes and suffixes all have a ratio of N that
  * is less than or equal to RATIO_TOO_MANY_N
+ *
+ * The parameters required_start and required_end give the positions
+ * of the required sequence that must not be cut out.
+ * @post start_pos <= required_start && length >= required_length
  */
-void trimSequence(string &sequence, size_t &start_pos, size_t &length);
-
+void trimSequence(string &sequence, size_t &start_pos, size_t &length,
+                  size_t required_start=string::npos, size_t required_length=0);
 
 const Sequence NULL_SEQUENCE = create_sequence("", "", "NULL", "");
 

algo/tests/unit-tests/testTools.cpp

@@ -3,6 +3,7 @@
 #include "tests.h"
 #include <stdexcept>
 #include <vector>
+#include <tuple>
 
 void testOnlineBioReader1() {
   OnlineBioReader *fa = OnlineBioReaderFactory::create("data/test1.fa");
@@ -470,6 +471,28 @@ void testTrimSequence() {
                "got " << trimmed << " instead of " << seq_pair.second << " (original sequence: " << seq_pair.first
       << ")");
     }
+
+    // Test the last parameters
+
+    //                       0        0  1          2  2 
+    //                       0        9  2          3  6
+    string representative = "TTTTTTTTTNNNNCCCCCCCCCCNNNNAAAAAAAAA";
+
+    list <std::tuple<size_t, size_t, string> > required_params = 
+      { std::make_tuple(13, 10, "CCCCCCCCCC"),
+        std::make_tuple(13, 11, "CCCCCCCCCCN"),
+        std::make_tuple(12, 10, "NCCCCCCCCC"),
+        std::make_tuple(12, 11, "NCCCCCCCCCC"),
+        std::make_tuple(12, 12, "NCCCCCCCCCCN"),
+        std::make_tuple(11, 14, "NNCCCCCCCCCCNN")};
+
+    for (auto ex: required_params) {
+      start = 0;
+      length = representative.length();
+      trimSequence(representative, start, length, std::get<0>(ex), std::get<1>(ex));
+      trimmed = representative.substr(start, length);
+      TAP_TEST_EQUAL(trimmed, std::get<2>(ex), TEST_TRIM_SEQUENCE, " required_start = " << std::get<0>(ex) << ", required_length = " << std::get<1>(ex));
+    }
 }
 
 /*