Checks for forcing sorting and indexing of bam files.

src/portculis.cc

@@ -59,6 +59,7 @@ int main(int argc, char *argv[]) {
         string genomeFile;
         string outputPrefix;
         bool collectiveMode;
+        bool forcePrep;
         uint16_t threads;
         bool disableThreadedIO;
         uint32_t gap_size;
@@ -69,7 +70,7 @@ int main(int argc, char *argv[]) {
         generic_options.add_options()
                 ("output_prefix,o", po::value<string>(&outputPrefix)->default_value(DEFAULT_OUTPUT_PREFIX), "Path prefix for files generated by this program.")
                 ("collective,c", po::value<bool>(&collectiveMode)->default_value(false), "Whether to treat all provided bam files as one large merged bam file or to handle each separately.") 
-
+                ("force_prep,fp", po::value<bool>(&forcePrep)->default_value(false), "Whether to force preparation (sorting and indexing) of the input bam files.  By default, portculis only sorts bam files if the header does not contain a sorted (coordinate) flag.  Also, it only indexes if it cannot find a bam file with a bamtools bti extension suffix.")
                 ("threads,t", po::value<uint16_t>(&threads)->default_value(DEFAULT_THREADS), "The number of threads to use.")
                 ("disable_threaded_io", po::value<bool>(&disableThreadedIO)->default_value(false), "Whether to acquire data in parallel to processing.")
                 ("gap,g", po::value<uint32_t>(&gap_size)->default_value(DEFAULT_GAP_SIZE), "The minimum gap size between adjacent alignments that determines whether or not we can chunk the data at this point.")
@@ -173,6 +174,7 @@ int main(int argc, char *argv[]) {
 
         // In collective mode make sure all bam files are merged and then sorted 
         // and indexed.  Create a vector of the index files.
+        // TODO: Might be possible to replace these with BamReaders instead!
         vector<string> indexFiles;
         vector<string> modBamFiles;
         if (collectiveMode) {
@@ -194,16 +196,18 @@ int main(int argc, char *argv[]) {
                 string bamFileToIndex = bamFile;
 
                 // Sort the BAM file if necessary
-                if (!portculis::isSortedBam(bamFile)) {
+                if (!portculis::isSortedBam(bamFile) || forcePrep) {
                     string sortedBam = bamFile + ".sorted.bam";
                     portculis::sortBam(bamFile, sortedBam, false);
                     bamFileToIndex = sortedBam;
                 }
 
                 string indexedBam = bamFile + ".sorted.bam.bti";
 
-                // Index the BAM file.
-                portculis::indexBam(bamFileToIndex, indexedBam);
+                // Index the BAM file if necessary.
+                if (!boost::filesystem::exists(indexedBam) || forcePrep) {
+                    portculis::indexBam(bamFileToIndex, indexedBam);
+                }
                 indexFiles.push_back(indexedBam);
                 modBamFiles.push_back(bamFileToIndex);
             }