Refactor alternative matrix parameter

inab · Oct 25, 2023 · 734ffa9 · 734ffa9
1 parent 3bd6e3e
commit 734ffa9
Show file tree

Hide file tree

Showing 10 changed files with 44 additions and 95 deletions.
diff --git a/Documentation/00_mainPage.dox b/Documentation/00_mainPage.dox
@@ -180,9 +180,8 @@ For a complete list please see the User Guide or visit http://trimal.cgenomics.o
         Split input coding sequences up to first stop codon appearance
 
 <b>Trimming Parameters</b>
-    --alternative_matrix degenerated_nt_identity
+    --degenerated_nt_identity
         Specify the degenerated nt identity matrix as the similarity matrix to use.
-        If a matrix is not specified, the best suited among a set will be selected.
     -matrix <inpufile>
         Input file for user-defined similarity matrix (default is Blosum62).
     -block <n>

diff --git a/Documentation/ClassDocumentation/trimAlManager/trimAlManager-variables.dox b/Documentation/ClassDocumentation/trimAlManager/trimAlManager-variables.dox
@@ -206,7 +206,7 @@ If the counter is bigger than 1, we are trying to use multiple automatic methods
 */
 
 /**
-\var int trimAlManager::alternative_matrix
+\var int trimAlManager::degenerated_nt_identity
 \brief Int that represents which alternative matrix to use, where '-1' means no alternative matrix, and '1' means degenerated_nt_identity
 */
 

diff --git a/docs/source/usage.rst b/docs/source/usage.rst
@@ -125,10 +125,9 @@ Backtranslation Options
 
 Trimming Parameters
 =======================
-.. option:: --alternative_matrix <degenerated_nt_identity>
+.. option:: --degenerated_nt_identity
 
         Specify the degenerated nt identity matrix as the similarity matrix to use.
-        If a matrix is not specified, the best suited among a set will be selected.
 
 .. option:: -matrix <inputfile>
 

diff --git a/include/RawText/legacyMenu.txt b/include/RawText/legacyMenu.txt
@@ -151,9 +151,8 @@ Common options (for a complete list please see the User Guide or visit http://tr
     -ignoreFilter            Ignore vcf-filtered variants in VCF. Only valid in combination with -vcf.
                                 Still applies min-quality and min-coverage when provided
 
-    --alternative_matrix degenerated_nt_identity
+    --degenerated_nt_identity
                              Specify the degenerated nt identity matrix as the similarity matrix to use.
-                                If a matrix is not specified, the best suited among a set will be selected.
 
 Legacy Options
     These options are included for back-compatibility with older versions of trimAl.

diff --git a/include/RawText/menu.txt b/include/RawText/menu.txt
@@ -106,9 +106,8 @@ For a complete list please see the User Guide or visit http://trimal.cgenomics.o
         Split input coding sequences up to first stop codon appearance
 
 [b][u]Trimming Parameters[r]
-    [b]--alternative_matrix degenerated_nt_identity[r]
+    [b]--degenerated_nt_identity[r]
         Specify the degenerated nt identity matrix as the similarity matrix to use.
-        If a matrix is not specified, the best suited among a set will be selected.
     [b]-matrix <inpufile>[r]
         Input file for user-defined similarity matrix (default is Blosum62).
     [b]-block <n>[r]

diff --git a/include/Statistics/similarityMatrix.h b/include/Statistics/similarityMatrix.h
@@ -101,14 +101,9 @@ namespace statistics {
         void defaultNTDegeneratedSimMatrix();
 
         /**
-         * \brief Method to load alternative similarity matrices also included on the suite.
-         * Currently, only one type of alternative matrix is available: \n
-         * \b matrix_code: 1 \b datatype SequenceTypes::AA
-         * \param matrix_code ID of the matrix
-         * \param datatype Numberical representation of the data type.
-         *  See #SequenceTypes
+         * \brief Method to load alternative DEG NT similarity matrix
         */
-        void alternativeSimilarityMatrices(int matrix_code, int datatype);
+        void alternativeNTDegeneratedSimMatrix();
 
         /**
          * \brief Method to get the similarity distance between two residues, A and B\n

diff --git a/include/trimalManager.h b/include/trimalManager.h
@@ -118,7 +118,7 @@ class trimAlManager
         clusters                = -1
         ,
         automatedMethodCount    = -1,
-        alternative_matrix      = -1,
+        degenerated_nt_identity = -1,
         gapAbsoluteThreshold    = -1,
 
         *delColumns         = nullptr,

diff --git a/source/Statistics/similarityMatrix.cpp b/source/Statistics/similarityMatrix.cpp
@@ -350,62 +350,31 @@ namespace statistics {
         }
     }
 
-    void similarityMatrix::alternativeSimilarityMatrices(int matrix_code, \
-        int datatype) {
-        int i, j, k;
+    void similarityMatrix::alternativeNTDegeneratedSimMatrix() {
         float sum;
 
-        // Allocate memory depending on the input datatype
-        switch (datatype) {
-            case SequenceTypes::AA:
-                memoryAllocation(20);
-                break;
-            case SequenceTypes::DNA:
-            case SequenceTypes::RNA:
-                memoryAllocation(5);
-                break;
-            case SequenceTypes::DNA | SequenceTypes::DEG:
-            case SequenceTypes::RNA | SequenceTypes::DEG:
-                memoryAllocation(15);
-                break;
-        }
-
-        for (i = 0; i < TAMABC; i++)
+        memoryAllocation(15);
+        for (int i = 0; i < TAMABC; i++) {
             vhash[i] = -1;
+        }
 
         // We create the hashing vector taking into account the input datatype
-        for (i = 0; i < numPositions; i++) {
-            switch (datatype) {
-                case SequenceTypes::AA:
-                    vhash[listAASym[i] - 'A'] = i;
-                    break;
-                case SequenceTypes::DNA:
-                case SequenceTypes::RNA:
-                    vhash[listNTSym[i] - 'A'] = i;
-                    break;
-                case SequenceTypes::DNA | SequenceTypes::DEG:
-                case SequenceTypes::RNA | SequenceTypes::DEG:
-                    vhash[listNTDegenerateSym[i] - 'A'] = i;
-                    break;
-            }
+        for (int i = 0; i < numPositions; i++) {
+            vhash[listNTDegenerateSym[i] - 'A'] = i;
         }
 
         // Working similarity matrix is set depending on the pre loaded matrices
-        for (i = 0; i < numPositions; i++) {
-            for (j = 0; j < numPositions; j++) {
-                switch (matrix_code) {
-                    case 1:
-                        simMat[i][j] = alternative_1_NTDegeneratedMatrix[i][j];
-                        break;
-                }
+        for (int i = 0; i < numPositions; i++) {
+            for (int j = 0; j < numPositions; j++) {
+                simMat[i][j] = alternative_1_NTDegeneratedMatrix[i][j];
             }
         }
 
         // Calculate the distances between residues based on Euclidean distance
-        for (j = 0; j < numPositions; j++) {
-            for (i = 0; i < numPositions; i++) {
+        for (int j = 0; j < numPositions; j++) {
+            for (int i = 0; i < numPositions; i++) {
                 if ((i != j) && (distMat[i][j] == 0.0)) {
-                    for (k = 0, sum = 0; k < numPositions; k++)
+                    for (int k = 0, sum = 0; k < numPositions; k++)
                         sum += ((simMat[k][j] - simMat[k][i]) * (simMat[k][j] - simMat[k][i]));
                     sum = (float) sqrt(sum);
                     distMat[i][j] = sum;

diff --git a/source/trimalManager.cpp b/source/trimalManager.cpp
@@ -444,17 +444,9 @@ int trimAlManager::parseArguments(int argc, char **argv) {
         matrixFile = new char[argumentLength + 1];
         strcpy(matrixFile, argv[*i]);
         return Recognized;
-    } else if (!strcmp(argv[*i], "--alternative_matrix") && ((*i) + 1 != *argc) && (alternative_matrix == -1)) {
-        (*i)++;
-        if (!strcmp(argv[*i], "degenerated_nt_identity"))
-        {
-            alternative_matrix = 1;
-            return Recognized;
-        }
-        else {
-            debug.report(ErrorCode::AlternativeMatrixNotRecognized, argv[*i]);
-            appearErrors = true;
-        }
+    } else if (!strcmp(argv[*i], "--degenerated_nt_identity") && ((*i) + 1 != *argc) && (degenerated_nt_identity == -1)) {
+        degenerated_nt_identity = 1;
+        return Recognized;
     }
     return NotRecognized;
 }
@@ -1376,7 +1368,7 @@ inline bool trimAlManager::check_absolute_gap_theshold() {
 }
 
 /**inline**/ bool trimAlManager::check_similarity_matrix() {
-    if ((matrixFile != nullptr || alternative_matrix != -1) && (!appearErrors)) {
+    if ((matrixFile != nullptr || degenerated_nt_identity != -1) && (!appearErrors)) {
         if ((!strict) && (!strictplus) && (!automated1) && (similarityThreshold == -1) && (!ssc) && (!sst)) {
             debug.report(ErrorCode::MatrixGivenWithNoMethodToUseIt);
             appearErrors = true;
@@ -1967,33 +1959,37 @@ int trimAlManager::perform() {
     if ((strict) || (strictplus) || (automated1) || (similarityThreshold != -1.0) || (ssc == 1) || (sst == 1)) {
         similMatrix = new statistics::similarityMatrix();
 
-    // Load Matrix
-        if (matrixFile != nullptr)
+        if (matrixFile != nullptr) {
             similMatrix->loadSimMatrix(matrixFile);
+        } else if (degenerated_nt_identity != -1) {
+            int alignDataType = origAlig->getAlignmentType();
+            if (alignDataType == (SequenceTypes::DNA | SequenceTypes::DEG) ||
+                alignDataType == (SequenceTypes::RNA | SequenceTypes::DEG)) {
+              debug.report(ErrorCode::ImpossibleToProcessMatrix);
+              return false;
+            }
 
-    // Alternative Default Matrix
-        else if (alternative_matrix != -1) {
-            similMatrix->alternativeSimilarityMatrices(alternative_matrix, origAlig->getAlignmentType());
-        }
-
-    // Default Matrices
-        else {
+            similMatrix->alternativeNTDegeneratedSimMatrix();
+        } else {
             int alignDataType = origAlig->getAlignmentType();
-            if (alignDataType & SequenceTypes::AA)
+            if (alignDataType == SequenceTypes::AA) {
                 similMatrix->defaultAASimMatrix();
-            else if ((alignDataType == SequenceTypes::DNA) || (alignDataType == SequenceTypes::RNA))
+            } else if (alignDataType == (SequenceTypes::AA | SequenceTypes::DEG)) {
+                similMatrix->defaultAASimMatrix();
+            } else if ((alignDataType == SequenceTypes::DNA) || (alignDataType == SequenceTypes::RNA)) {
                 similMatrix->defaultNTSimMatrix();
-            else if ((alignDataType == (SequenceTypes::DNA | SequenceTypes::DEG)) ||
-                     (alignDataType == (SequenceTypes::RNA | SequenceTypes::DEG)))
-                similMatrix->defaultNTDegeneratedSimMatrix();
+            } else if ((alignDataType == (SequenceTypes::DNA | SequenceTypes::DEG)) ||
+                     (alignDataType == (SequenceTypes::RNA | SequenceTypes::DEG))) {
+                        similMatrix->defaultNTDegeneratedSimMatrix();
+            }
         }
 
-    // Check if Matrix has been loaded
         if (!origAlig->Statistics->setSimilarityMatrix(similMatrix)) {
             debug.report(ErrorCode::ImpossibleToProcessMatrix);
             return false;
         }
     }
+
     return true;
 }
 

diff --git a/tests/source/trimAlManager/parseArguments/MatrixArguments.cpp b/tests/source/trimAlManager/parseArguments/MatrixArguments.cpp
@@ -131,7 +131,7 @@ TEST_CASE("Matrix", "[manager][arguments][matrix]") {
 
     GIVEN("Alternative matrix argument")
     {
-        args.push_back("--alternative_matrix");
+        args.push_back("--degenerated_nt_identity");
         test_arguments(
                 args,
                 manager,
@@ -140,13 +140,6 @@ TEST_CASE("Matrix", "[manager][arguments][matrix]") {
 
         AND_GIVEN("Correct value input")
         {
-            args.push_back("degenerated_nt_identity");
-            test_arguments(
-                    args,
-                    manager,
-                    true, trimAlManager::argumentReport::Errored,
-                    true, true);
-
             GIVEN("Input file")
             {
                 args.push_back("-in");