Updated cut-off scheme

extractor/debug.cc

@@ -8,8 +8,8 @@
 // FILE INFORMATION:
 //   File:     debug.cc
 //   Author:   Jonathan K. Vis
-//   Revision: 2.1.0
-//   Date:     2014/08/06
+//   Revision: 2.1.1
+//   Date:     2014/08/21
 // *******************************************************************
 // DESCRIPTION:
 //   This source can be used to debug the Extractor library within
@@ -46,7 +46,8 @@ int main(int argc, char* argv[])
   size_t const reference_length = ftell(file);
   rewind(file);
   char_t* reference = new char_t[reference_length];
-  fread(reference, sizeof(char_t), reference_length, file);
+  size_t const ref_length = fread(reference, sizeof(char_t), reference_length, file);
+  static_cast<void>(ref_length);
   fclose(file);
 
   file = fopen(argv[2], "r");
@@ -60,7 +61,8 @@ int main(int argc, char* argv[])
   size_t const sample_length = ftell(file);
   rewind(file);
   char_t* sample = new char_t[sample_length];
-  fread(sample, sizeof(char_t), sample_length, file);
+  size_t const alt_length = fread(sample, sizeof(char_t), sample_length, file);
+  static_cast<void>(alt_length);
   fclose(file);
 
 

extractor/extractor.cc

@@ -8,8 +8,8 @@
 // FILE INFORMATION:
 //   File:     extractor.cc (depends on extractor.h)
 //   Author:   Jonathan K. Vis
-//   Revision: 2.1.2
-//   Date:     2014/08/13
+//   Revision: 2.1.4
+//   Date:     2014/08/21
 // *******************************************************************
 // DESCRIPTION:
 //   This library can be used to generete HGVS variant descriptions as
@@ -217,8 +217,9 @@ size_t extractor(std::vector<Variant> &variant,
 
   // Calculate the LCS (possibly in reverse complement) of the two
   // strings.
+  size_t const cut_off = reference_length < THRESHOLD_CUT_OFF ? 1 : EXTRACTION_CUT_OFF;
   std::vector<Substring> substring;
-  size_t const length = LCS(substring, reference, complement, reference_start, reference_end, sample, sample_start, sample_end);
+  size_t const length = LCS(substring, reference, complement, reference_start, reference_end, sample, sample_start, sample_end, cut_off);
 
 
   // No LCS found: this is a transposition or a deletion/insertion.
@@ -456,8 +457,9 @@ size_t extractor_transposition(std::vector<Variant> &variant,
 
 
   // Extract the LCS (from the whole reference string).
+  size_t const cut_off = static_cast<size_t>(TRANSPOSITION_CUT_OFF * static_cast<double>(sample_length));
   std::vector<Substring> substring;
-  size_t const length = LCS(substring, reference, complement, 0, global_reference_length, sample, sample_start, sample_end);
+  size_t const length = LCS(substring, reference, complement, 0, global_reference_length, sample, sample_start, sample_end, cut_off);
 
 
   // No LCS found: this is a deletion/insertion.
@@ -557,18 +559,12 @@ size_t LCS(std::vector<Substring> &substring,
            size_t const            reference_end,
            char_t const* const     sample,
            size_t const            sample_start,
-           size_t const            sample_end)
+           size_t const            sample_end,
+           size_t const            cut_off)
 {
   size_t const reference_length = reference_end - reference_start;
   size_t const sample_length = sample_end - sample_start;
 
-  // Always fully explore strings smaller than this threshold.
-  static size_t const THRESHOLD = 16000;
-
-  // A dynamic cut-off for unevenly matched string lengths.
-  double const a = reference_length >= sample_length ? reference_length : sample_length;
-  double const b = reference_length >= sample_length ? sample_length : reference_length;
-  size_t const cut_off = (reference_length > THRESHOLD ? ceil((1.0 - b / (a + 0.1 * b)) * b) / 8 : 0) + 1;
 
   // The initial k.
   size_t k = reference_length > sample_length ? sample_length / 4 : reference_length / 4;

extractor/extractor.h

@@ -8,8 +8,8 @@
 // FILE INFORMATION:
 //   File:     extractor.h (implemented in extractor.cc)
 //   Author:   Jonathan K. Vis
-//   Revision: 2.1.2
-//   Date:     2014/08/13
+//   Revision: 2.1.4
+//   Date:     2014/08/21
 // *******************************************************************
 // DESCRIPTION:
 //   This library can be used to generate HGVS variant descriptions as
@@ -34,7 +34,7 @@ namespace mutalyzer
 {
 
 // Version string for run-time identification.
-static char const* const VERSION = "2.1.2";
+static char const* const VERSION = "2.1.4";
 
 
 // The character type used for all strings. For now it should just be
@@ -90,6 +90,15 @@ static size_t const WEIGHT_SEPARATOR          = 1; // i.e., _, [, ], ;
 static size_t const WEIGHT_SUBSTITUTION       = 1; // i.e., >
 
 
+// Cut-off constants. For normal extraction use the threshold to
+// specify the maximum reference length without any cut-off. Otherwise
+// the extraction cut-off is used.
+// The for transpositions is set at a fraction of the sample length.
+static size_t const THRESHOLD_CUT_OFF     = 16000;
+static size_t const EXTRACTION_CUT_OFF    =   500;
+static double const TRANSPOSITION_CUT_OFF =   0.3;
+
+
 // This global variable is used to have access to the whole reference
 // string at any point in the extraction process. Commonly used in
 // transposition extraction.
@@ -317,6 +326,7 @@ struct Substring
 //   @arg sample: sample string
 //   @arg sample_start: starting position in the sample string
 //   @arg sample_end: ending position in the sample string
+//   @arg cut_off: optional cut-off value for the k in LCS_k
 //   @return: length of the LCS
 // *******************************************************************
 size_t LCS(std::vector<Substring> &substring,
@@ -326,7 +336,8 @@ size_t LCS(std::vector<Substring> &substring,
            size_t const            reference_end,
            char_t const* const     sample,
            size_t const            sample_start,
-           size_t const            sample_end);
+           size_t const            sample_end,
+           size_t const            cut_off = 1);
 
 // *******************************************************************
 // LCS_1 function

extractor/extractor.i

@@ -8,8 +8,8 @@
 // FILE INFORMATION:
 //   File:     extractor.i (SWIG interface file)
 //   Author:   Jonathan K. Vis
-//   Revision: 2.1.2
-//   Date:     2014/08/13
+//   Revision: 2.1.4
+//   Date:     2014/08/21
 // *******************************************************************
 // DESCRIPTION:
 //   Defines the SWIG interface for the Extractor library for use in
@@ -32,7 +32,7 @@ namespace mutalyzer
 {
 
 // Version string for run-time identification.
-static char const* const VERSION = "2.1.2";
+static char const* const VERSION = "2.1.4";
 
 // The character type used for all strings. For now it should just be
 // a char.
@@ -85,8 +85,6 @@ static size_t const WEIGHT_SUBSTITUTION       = 1; // i.e., >
 //                       sample string
 //   @member type: type of the variant described using the
 //                 constants above
-//   @member weight: weight of the variant according to the weight
-//                   constants above (used internally)
 //   @member transposition_start: starting position of a transposition
 //                                withing the reference string
 //   @member transposition_end: ending position of a transposition
@@ -99,7 +97,6 @@ struct Variant
   size_t       sample_start;
   size_t       sample_end;
   unsigned int type;
-  size_t       weight;
   size_t       transposition_start;
   size_t       transposition_end;
 }; // Variant