adjusted parameters

removed pacBio and uon presetting

Author: MarkusRainerSchmidt

Makefile

@@ -26,8 +26,14 @@ TARGET_OBJ= \
 
 # flags
 CC=gcc
-CCFLAGS= -Wall -Werror -fPIC -std=c++11 -O3 -g
-CFLAGS= -Wall -Werror -fPIC -O3 -g
+# use avx instead of sse
+ifeq ($(NO_SSE), 1)
+	CCFLAGS= -Wall -Werror -fPIC -std=c++11 -O3 -g
+	CFLAGS= -Wall -Werror -fPIC -O3 -g
+else
+	CCFLAGS= -Wall -Werror -fPIC -std=c++11 -O3 -g -msse4.1
+	CFLAGS= -Wall -Werror -fPIC -O3 -g -msse4.1
+endif
 LDFLAGS= -std=c++11
 LDLIBS= -L$(LIBGABA_HOME) -lm -lpthread -lstdc++ -lgaba
 INCLUDES= -isystem$(LIBGABA_HOME)/ -Iinc
@@ -59,10 +65,6 @@ else
 	MA_REQUIREMENT += $(TARGET_OBJ) $(LIBGABA_HOME)/libgaba.a
 endif
 
-# use avx instead of sse
-ifeq ($(WITH_AVX2), 1)
-	CCFLAGS += -mavx2
-endif
 
 # compile with postgre support enabled
 ifeq ($(WITH_POSTGRES), 1)

README.md

@@ -62,7 +62,12 @@ Possible Makefile switches:
     *Tested using boost 1.65.1 (with the Boost.Python library) and Python3.*
 
 - DEBUG=1\
-    Compiles MA without optimizations and with assertions enabled. Further, there are multiple self-checks during runtime. The debug mode is primarily for bug hunting.
+    Compiles MA without optimizations and with assertions enabled. Further, there are multiple 
+    self-checks during runtime. The debug mode is primarily for bug hunting.
+
+- NO_SSE=1\
+    Compiles MA without SIMD (sse4.1) instructions. This will result in much worse runtimes and 
+    should therefore only be enabled if the underlying hardware does not support sse4.1.
 
 ### First Steps ...
 Test your installation with:
@@ -107,11 +112,10 @@ Available presettings:
      -m, --mode <str>              Operation mode for MA. (default is 'fast')
                                    fast:
                                        Best compromise between performance and accuracy.
+                                       Recommended for PacBio reads.
                                    acc:
                                        Better accuracy than in fast mode but worse runtimes.
-                                       Particularly effective for short reads.  
-                                   pacBio:
-                                       Operation mode for PacBio sequencer reads.
+                                       Particularly effective for short reads.
 
 Alignment options:
     -o, --out <fname>              Filename used for SAM file output. Default output stream is
@@ -165,7 +169,7 @@ Advanced options:
                                    Default is 10000000.
         --minSecToPrimRatio <num>  Limit output of secondary alignments to alignments with: 
                                    score-secondary-alignment >= <num> * score-primary-alignment.
-                                   Default is 0.75.
+                                   Default is 0.25.
         --maxOverlapSupp <num>     A secondary alignment becomes supplementary, if it overlaps less
                                    than <num> percent with the primary alignment.
                                    Default is 0.1.

inc/module/binarySeeding.h

@@ -474,7 +474,7 @@ namespace libMA
             );
 
     public:
-        size_t uiUONAccMaxSize;
+        size_t uiUONAccMaxSize = defaults::uiUONAccMaxSize;
         /**
          * @brief Initialize a BinarySeeding Module
          * @details
@@ -486,8 +486,7 @@ namespace libMA
          */
         BinarySeeding()
                 :
-            bLrExtension(defaults::sSeedSet == "maxSpan"),
-            uiUONAccMaxSize(defaults::uiUONAccMaxSize)
+            bLrExtension(defaults::sSeedSet == "maxSpan")
         {}//constructor
 
         std::shared_ptr<Container> EXPORTED execute(std::shared_ptr<ContainerVector> vpInput);

inc/util/default_parameters.h

@@ -76,30 +76,13 @@ namespace libMA
             sSeedSet = "maxSpan";
         }// function
 
-        inline void configurePacBio()
-        {
-            sParameterSet = "pacBio";
-            sSeedSet = "maxSpan";
-            bDisableHeuristics = true;
-            dMaxOverlapSupplementary = 0.25;
-            fMinSecScoreRatio = 0.85;
-            uiReportN = 3;
-            uiPadding = 1000;
-            //bDisableGapCostEstimationCutting = true;
-        }// function
-
-        inline void configureUON()
-        {
-            sParameterSet = "uon";
-            sSeedSet = "maxSpan";
-            bDisableHeuristics = true;
-            dMaxOverlapSupplementary = 0.25;
-            fMinSecScoreRatio = 0.85;
-            uiReportN = 3;
-            uiPadding = 1000;
-            uiUONAccMaxSize = 800;
-            //bDisableGapCostEstimationCutting = true;
-        }// function
+        //inline void configurePacBio()
+        //{
+        //    sParameterSet = "pacBio";
+        //    sSeedSet = "maxSpan";
+        //    bDisableHeuristics = true;
+        //    uiReportN = 3;
+        //}// function
 
 #ifdef WITH_PYTHON
         void exportDefaults();

src/cmdMa.cpp

@@ -51,11 +51,10 @@ const std::string sHelp =
 "\n     -m, --mode [str]              Operation mode for MA. (default is 'fast')"
 "\n                                   fast:"
 "\n                                       Best compromise between performance and accuracy."
+"\n                                       Recommended for PacBio reads."
 "\n                                   acc:"
 "\n                                       Better accuracy than in fast mode but worse runtimes."
-"\n                                       Particularly effective for short reads.  "
-"\n                                   pacBio:"
-"\n                                       Operation mode for PacBio sequencer reads."
+"\n                                       Particularly effective for short reads."
 "\n"
 "\nAlignments options:"
 "\n    -o, --out <fname>              Filename used for SAM file output. Default output stream is"
@@ -116,7 +115,7 @@ const std::string sHelp =
 "\n                                   Default is 10,000,000."
 "\n        --minSecToPrimRatio <num>  Limit output of secondary alignments to alignments with:"
 "\n                                   score-secondary-alignment >= <num> * score-primary-alignment."
-"\n                                   Default is 0.75."
+"\n                                   Default is 0.25."
 "\n        --maxOverlapSupp <num>     A secondary alignment becomes supplementary, if it overlaps"
 "\n                                   less than <num> percent with the primary alignment."
 "\n                                   Default is 0.1."
@@ -163,10 +162,6 @@ int main(int argc, char* argv[])
                     defaults::configureAccurate();
                 if(strcmp(argv[i+1], "fast") == 0)
                     defaults::configureFast();
-                if(strcmp(argv[i+1], "pacBio") == 0)
-                    defaults::configurePacBio();
-                if(strcmp(argv[i+1], "uon") == 0)
-                    defaults::configureUON();
             }// if
 
         options.add_options("Alignment options (requires -a)")

src/util/default_parameters.cpp

@@ -23,7 +23,7 @@ namespace libMA
         size_t uiMaxEqualScoreLookahead = 3;     // lookahead distance for short queries
         size_t uiSwitchQLen = 800;       // q len to switch between break criteria
         uint64_t uiMaxGapArea = 10000;     // break alignments in harmonization if gap is larger
-        uint64_t uiPadding = 300;          // padding for DP
+        uint64_t uiPadding = 1000;          // padding for DP
         size_t uiSoCWidth = 0;          // set a fixed SoC width; 0 = use the formula
         bool bFindMode = false;        // true: don't do DP
         bool bOptimisticGapEstimation = true;  // how to estimate gap costs in harmonization
@@ -44,7 +44,7 @@ namespace libMA
         size_t uiGenomeSizeDisable = 10000000;
         bool bDisableHeuristics = false; // disable all heuristics in the harmonization
         // only output secondary alignments with a score larger than this * score of prim. alignment
-        float fMinSecScoreRatio = .75;
+        float fMinSecScoreRatio = .25;
         // can only be accessed by pacBio presetting at the moment...
         bool bDisableGapCostEstimationCutting = false; // do not remove seeds that are far away to generate alignments with positive score