Implemented --outBAMsortingBinsN option to control the number of sort…

…ing bins. Icnreasing this number reduces the amount of RAM required for sorting.

CHANGES.md

@@ -1,3 +1,4 @@
+* Implemented --outBAMsortingBinsN option to control the number of sorting bins. Icnreasing this number reduces the amount of RAM required for sorting.
 * Implemented --waspOutputMode option for filtering allele specific alignments. This is re-implemenation of the original WASP algorithm by Bryce van de Geijn, Graham McVicker, Yoav Gilad & Jonathan K Pritchard. Please cite the original WASP paper: Nature Methods 12, 1061–1063 (2015), https://www.nature.com/articles/nmeth.3582 .
 
 * Fixed a bug in chimeric detection code which sometimes led to uninitialized memory access. The chimeric output may change for a very small number of reads.

source/BAMbinSortUnmapped.cpp

@@ -12,11 +12,11 @@ void BAMbinSortUnmapped(uint32 iBin, uint nThreads, string dirBAMsort, BGZF *bgz
     vector<string> bamInFile;
     std::map <uint,uint> startPos;
 
-    for (uint it=0; it<nThreads; it++) {//initialize
+    for (uint it=0; it<nThreads; it++) {//files from all threads, and BySJout
         bamInFile.push_back(dirBAMsort+to_string(it)+"/"+to_string((uint) iBin));
         bamInFile.push_back(dirBAMsort+to_string(it)+"/"+to_string((uint) iBin)+".BySJout");
     };
-    vector<uint32> bamSize(bamInFile.size(),0);
+    vector<uint32> bamSize(bamInFile.size(),0);//record sizes
 
     //allocate arrays
     char **bamIn=new char* [bamInFile.size()];
@@ -25,13 +25,13 @@ void BAMbinSortUnmapped(uint32 iBin, uint nThreads, string dirBAMsort, BGZF *bgz
     for (uint it=0; it<bamInFile.size(); it++) {//initialize
         bamIn[it] = new char [BAMoutput_oneAlignMaxBytes];
 
-        bamInStream[it].open(bamInFile.at(it).c_str());
+        bamInStream[it].open(bamInFile.at(it).c_str());//opean all files
 
-        bamInStream[it].read(bamIn[it],sizeof(int32));//read record size
+        bamInStream[it].read(bamIn[it],sizeof(int32));//read BAM record size
         if (bamInStream[it].good()) {
-            bamSize[it]=((*(uint32*)bamIn[it])+sizeof(int32));
+            bamSize[it]=((*(uint32*)bamIn[it])+sizeof(int32));//true record size +=4 (4 bytes for uint-iRead)
             bamInStream[it].read(bamIn[it]+sizeof(int32),bamSize.at(it)-sizeof(int32)+sizeof(uint));//read the rest of the record, including last uint = iRead
-            startPos[*(uint*)(bamIn[it]+bamSize.at(it))]=it;
+            startPos[*(uint*)(bamIn[it]+bamSize.at(it))]=it;//startPos[iRead]=it : record the order of the files to output
         } else {//nothing to do here, file is empty, do not record it
         };
     };

source/Parameters.cpp

@@ -102,6 +102,7 @@ Parameters::Parameters() {//initalize parameters info
     parArray.push_back(new ParameterInfoScalar <string>     (-1, -1, "outSAMheaderCommentFile", &outSAMheaderCommentFile));
     parArray.push_back(new ParameterInfoScalar <int>        (-1, -1, "outBAMcompression", &outBAMcompression));
     parArray.push_back(new ParameterInfoScalar <int>        (-1, -1, "outBAMsortingThreadN", &outBAMsortingThreadN));
+    parArray.push_back(new ParameterInfoScalar <uint32>        (-1, -1, "outBAMsortingBinsN", &outBAMsortingBinsN));
     parArray.push_back(new ParameterInfoVector <string>     (-1, -1, "outSAMfilter", &outSAMfilter.mode));
     parArray.push_back(new ParameterInfoScalar <uint>     (-1, -1, "outSAMmultNmax", &outSAMmultNmax));
     parArray.push_back(new ParameterInfoScalar <uint>     (-1, -1, "outSAMattrIHstart", &outSAMattrIHstart));
@@ -601,7 +602,7 @@ void Parameters::inputParameters (int argInN, char* argIn[]) {//input parameters
                 } else {
                     outBAMsortingThreadNactual=outBAMsortingThreadN;
                 };
-                outBAMcoordNbins=max(outBAMsortingThreadNactual*3,10);
+                outBAMcoordNbins=max((uint32)outBAMsortingThreadNactual*3,outBAMsortingBinsN);
                 outBAMsortingBinStart= new uint64 [outBAMcoordNbins];
                 outBAMsortingBinStart[0]=1;//this initial value means that the bin sizes have not been determined yet
 

source/Parameters.h

@@ -135,6 +135,7 @@ class Parameters {
         vector <string> outSAMtype;
         bool outBAMunsorted, outBAMcoord, outSAMbool;
         uint32 outBAMcoordNbins;
+        uint32 outBAMsortingBinsN;//user-defined number of bins for sorting
         string outBAMsortTmpDir;
 
 //         string bamRemoveDuplicatesType;

source/STAR.cpp

@@ -375,7 +375,7 @@ int main(int argInN, char* argIn[]) {
 
         //check max size needed for sorting
         uint maxMem=0;
-        for (uint32 ibin=0; ibin<nBins-1; ibin++) {//check akk bins
+        for (uint32 ibin=0; ibin<nBins-1; ibin++) {//check all bins
             uint binS=0;
             for (int it=0; it<P.runThreadN; it++) {//collect sizes from threads
                 binS += RAchunk[it]->chunkOutBAMcoord->binTotalBytes[ibin]+24*RAchunk[it]->chunkOutBAMcoord->binTotalN[ibin];

source/parametersDefault

@@ -315,6 +315,9 @@ outBAMcompression       1
 outBAMsortingThreadN    0
     int: >=0: number of threads for BAM sorting. 0 will default to min(6,--runThreadN).
 
+outBAMsortingBinsN      50
+    int: >0:  number of genome bins fo coordinate-sorting
+
 ### BAM processing
 
 bamRemoveDuplicatesType  -