Adding options for changing cell barcode and UMI length to minnow

COMBINE-lab · Aug 14, 2019 · 6a031b5 · 6a031b5
1 parent 2a8134c
commit 6a031b5
Show file tree

Hide file tree

Showing 5 changed files with 27 additions and 10 deletions.
diff --git a/include/MinnowUtil.hpp b/include/MinnowUtil.hpp
@@ -174,17 +174,17 @@ namespace util{
     return static_cast<uint32_t>(std::exp(distr(eng)));
   }
 
-  inline std::vector<std::string> generateUMIList(){
+  inline std::vector<std::string> generateUMIList(const uint32_t UMI_LENGTH, const uint32_t POOL_SIZE){
     std::vector<std::string> whiteList ;
     std::cerr << "PRINTING DEBUG: POOL_SIZE " << POOL_SIZE << "\n\n" ;
     whiteList.resize(POOL_SIZE) ;
-    for(int i=0 ; i < POOL_SIZE; ++i){
+    for(size_t i=0 ; i < POOL_SIZE; ++i){
       whiteList[i] = genRandomSeq(UMI_LENGTH) ;
     }
     return whiteList ;
   }
 
-  inline std::vector<std::string> generateCBList(int numCells){
+  inline std::vector<std::string> generateCBList(int numCells, const uint32_t CB_LENGTH){
     std::vector<std::string> CBList;
     CBList.resize(numCells) ;
     for(int i=0 ; i < numCells; ++i){

diff --git a/include/ProgOpts.hpp b/include/ProgOpts.hpp
@@ -59,6 +59,9 @@ class SimulateOptions {
   double mutationProb{0.01};
   uint32_t numThreads{2};
   uint32_t numOfDoublets{0};
+
+  uint32_t CBLength{16} ;
+  uint32_t UMILength{10} ;
 };
 
 class EstimateOptions {

diff --git a/include/macros.hpp b/include/macros.hpp
@@ -13,10 +13,9 @@
 #define FRAGMENT_RANGE (FRAGMENT_END_DIST - FRAGMENT_START_DIST)
 
 
-#define CB_LENGTH 16
-#define UMI_LENGTH 10
-#define POOL_SIZE std::pow(4, UMI_LENGTH)
-//#define POOL_SIZE 1000
+//#define CB_LENGTH 16
+//#define UMI_LENGTH 10
+//#define POOL_SIZE std::pow(4, UMI_LENGTH)
 #define TENX_WHITELIST_LENGTH 737280
 
-#endif // MACROS_HEADER
+#endif // MACROS_HEADER
diff --git a/src/Minnow.cpp b/src/Minnow.cpp
@@ -74,6 +74,14 @@ int main(int argc, char* argv[]) {
     (option("--numMolFile") & 
     value("num mol file", simulateOpt.numMolFile)) %
     "Number of molecules generated from each cell",
+
+    (option("--CBLength") & 
+     value("Cell barcode length", simulateOpt.CBLength)) %
+    "Cell barcode length by default is 16",
+
+    (option("--UMILength") & 
+     value("Cell barcode length", simulateOpt.UMILength)) %
+    "Cell barcode length by default is 10",
 
     (option("--alevin-mode").set(simulateOpt.alevinMode, true)) %
     "The program would assume that the input matrix is obtained from Alevin",

diff --git a/src/MinnowSimulate.cpp b/src/MinnowSimulate.cpp
@@ -46,7 +46,9 @@ using SpinLockT = std::mutex;
 #define FRAGMENT_START_DIST 53 + READ_LEN // this is from the empirical limit e^4
 #define FRAGMENT_RANGE (FRAGMENT_END_DIST - FRAGMENT_START_DIST)
 
-
+uint32_t CB_LENGTH ;
+uint32_t UMI_LENGTH ;
+uint32_t POOL_SIZE ;
 
 #define _verbose(fmt, args...) fprintf(stderr, fmt, ##args)
 
@@ -1819,6 +1821,11 @@ void minnowSimulate(SimulateOptions& simOpts){
     auto& numOfCells = simOpts.numOfCells ;
     auto& numOfGenes = simOpts.numOfTranscripts ;
 
+    CB_LENGTH = simOpts.CBLength ;
+    UMI_LENGTH = simOpts.UMILength ;
+    POOL_SIZE = std::pow(4, UMI_LENGTH) ;
+
+
     //auto& numOfSampleCells = simOpts.sampleCells ;
     //auto& numOfPCRCycles = simOpts.numOfPCRCycles ;
     //auto& errorRate = simOpts.errorRate ;
@@ -2048,7 +2055,7 @@ void minnowSimulate(SimulateOptions& simOpts){
 
 
     // generate UMI list 
-    std::vector<std::string> UMIList = util::generateUMIList() ; 
+    std::vector<std::string> UMIList = util::generateUMIList(UMI_LENGTH, POOL_SIZE) ; 
 
     // If we usedbg none of other stuff needed