Added -scores option to tagBam

src/tagBam/tagBam.cpp

@@ -15,13 +15,14 @@
 // build
 TagBam::TagBam(const string &bamFile, const vector<string> &annoFileNames,
             const vector<string> &annoLables, const string &tag,
-            bool useNames, bool sameStrand, bool diffStrand, float overlapFraction):
+            bool useNames, bool useScores, bool sameStrand, bool diffStrand, float overlapFraction):
 
     _bamFile(bamFile),
     _annoFileNames(annoFileNames),
     _annoLabels(annoLables),
     _tag(tag),
     _useNames(useNames),
+    _useScores(useScores),
     _sameStrand(sameStrand),
     _diffStrand(diffStrand),
     _overlapFraction(overlapFraction)
@@ -92,15 +93,25 @@ void TagBam::Tag() {
                 // grab the current annotation file.
                 BedFile *anno = _annoFiles[i];
                 
-                if (!_useNames) {
+                if (!_useNames && !_useScores) {
                     // add the label for this annotation file to tag if there is overlap
                     if (anno->FindOneOrMoreOverlapsPerBin(a.chrom, a.start, a.end, a.strand, _sameStrand, _diffStrand, _overlapFraction)) 
                     {
                         annotations << _annoLabels[i] << ";";
                     }
                 }
+                // use the score field
+                else if (!_useNames && _useScores) {
+                    anno->FindOverlapsPerBin(a.chrom, a.start, a.end, a.strand, hits, _sameStrand, _diffStrand);
+                    for (size_t i = 0; i < hits.size(); ++i) {
+                        annotations << hits[i].score;
+                        if (i < hits.size() - 1) annotations << ",";
+                    }
+                    if (hits.size() > 0) annotations << ";";
+                    hits.clear();
+                }
                 // use the name field from the annotation files to populate tag
-                else {
+                else if (_useNames && !_useScores) {
                     anno->FindOverlapsPerBin(a.chrom, a.start, a.end, a.strand, hits, _sameStrand, _diffStrand);
                     for (size_t i = 0; i < hits.size(); ++i) {
                         annotations << hits[i].name;

src/tagBam/tagBam.h

@@ -40,8 +40,9 @@ public:
 
     // constructor
     TagBam(const string &bamFile, const vector<string> &annoFileNames,
-                const vector<string> &annoLabels, const string &tag,
-                bool useNames, bool sameStrand, bool diffStrand, float overlapFraction);
+                const vector<string> &annoLabels, const string &tag, 
+                bool useNames, bool useScores, bool sameStrand, 
+                bool diffStrand, float overlapFraction);
 
     // destructor
     ~TagBam(void);
@@ -55,6 +56,7 @@ private:
     string _bamFile;
     vector<string> _annoFileNames;
     vector<string> _annoLabels;
+        
     string _tag;
 
     // instance of a bed file class.
@@ -63,6 +65,7 @@ private:
 
     // should we use the name field from the annotation files?
     bool _useNames;
+    bool _useScores;
     
     // do we care about strandedness when tagging?
     bool _sameStrand;

src/tagBam/tagBamMain.cpp

@@ -34,14 +34,16 @@ int main(int argc, char* argv[]) {
     string tag = "YB";
 
     // parm flags
-    bool haveTag        = false;
-    bool haveFraction   = false;
-    bool useNames       = false;
-    bool sameStrand     = false;
-    bool diffStrand     = false;
-    bool haveBam        = false;
-    bool haveFiles      = false;
-    bool haveLabels     = false;
+    bool haveTag          = false;
+    bool haveFraction     = false;
+    bool useNames         = false;
+    bool useScores        = false;
+    bool sameStrand       = false;
+    bool diffStrand       = false;
+    bool haveBam          = false;
+    bool haveFiles        = false;
+    bool haveLabels       = false;
+
 
     // list of annotation files / names
     vector<string> inputFiles;
@@ -104,6 +106,9 @@ int main(int argc, char* argv[]) {
         else if (PARAMETER_CHECK("-names", 6, parameterLength)) {
             useNames = true;
         }
+        else if (PARAMETER_CHECK("-scores", 7, parameterLength)) {
+            useScores = true;
+        }
         else if (PARAMETER_CHECK("-s", 2, parameterLength)) {
             sameStrand = true;
         }
@@ -135,8 +140,8 @@ int main(int argc, char* argv[]) {
         cerr << endl << "*****" << endl << "*****ERROR: Need -i, -files" << endl << "*****" << endl;
         showHelp = true;
     }
-    if (!useNames && !haveLabels) {
-        cerr << endl << "*****" << endl << "*****ERROR: Need -labels or -names" << endl << "*****" << endl;
+    if (!useNames && !haveLabels && !useScores) {
+        cerr << endl << "*****" << endl << "*****ERROR: Need -labels or -names or -scores" << endl << "*****" << endl;
         showHelp = true;
     }
     if (sameStrand && diffStrand) {
@@ -147,13 +152,17 @@ int main(int argc, char* argv[]) {
         cerr << endl << "*****" << endl << "*****ERROR: Use -labels or -names, not both. " << endl << "*****" << endl;
         showHelp = true;
     }
+    if (useScores && useNames) {
+        cerr << endl << "*****" << endl << "*****ERROR: Use -scores or -names, not both. " << endl << "*****" << endl;
+        showHelp = true;
+    }
     if (haveTag && tag.size() > 2) {
         cerr << endl << "*****" << endl << "*****ERROR: Custom tags should be at most two characters per the SAM specification. " << endl << "*****" << endl;
         showHelp = true;
     }
 
     if (!showHelp) {
-        TagBam *ba = new TagBam(bamFile, inputFiles, inputLabels, tag, useNames, sameStrand, diffStrand, overlapFraction);
+        TagBam *ba = new TagBam(bamFile, inputFiles, inputLabels, tag, useNames, useScores, sameStrand, diffStrand, overlapFraction);
         ba->Tag();
         delete ba;
         return 0;
@@ -191,6 +200,9 @@ void ShowHelp(void) {
     
     cerr << "\t-names\t"        << "Use the name field from the annotation files to populate tags." << endl;
     cerr                        << "\t\tBy default, the -labels values are used." << endl << endl;
+
+    cerr << "\t-scores\t"    << "A list of 1-based columns for each annotation file" << endl;
+    cerr                        << "\t\tin which a color can be found." << endl << endl;
     
     
     exit(1);