hg38 epo

grenaud · May 9, 2022 · ccdb16e · ccdb16e
2 parents daed35e + 6d59fd6
commit ccdb16e
Show file tree

Hide file tree

Showing 91 changed files with 610 additions and 161 deletions.
diff --git a/.travis.yml b/.travis.yml
@@ -1,6 +1,6 @@
 language: C++
 
-before_script: make libgab/utils.o tabixpp/tabix.o && sed -i  "s/++11/++11\ -coverage/g" Makefile
+before_script: make libgab/libgab.a tabixpp/tabix.o && sed -i  "s/++11/++11\ -coverage/g" Makefile
 
 script: make test
 

diff --git a/ACF2BETASCAN.cpp b/ACF2BETASCAN.cpp
@@ -147,11 +147,14 @@ int ACF2BETASCAN::run(int argc, char *argv[]){
 
 	    if(arr->vectorAlleles->at(0).getAltCount() != 0 ){
 		if(arr->vectorAlleles->at(0).getRefCount() != 0 ){
-		    cerr<<"Cannot determine the root allele for "<<*arr<<endl;
-		    continue;
+		    char b=arr->vectorAlleles->at(0).generateRandomAlleleBias(arr->ref,arr->alt);
+		    rootIsRef = (b==arr->ref);
+		    //cerr<<"Cannot determine the root allele for "<<*arr<<endl;
+		    //continue;
 		    //return 1;      
-		}	
-		rootIsRef=false;
+		}else{	
+		    rootIsRef=false;
+		}
 	    }else{
 		rootIsRef=true;
 	    }
@@ -166,9 +169,11 @@ int ACF2BETASCAN::run(int argc, char *argv[]){
 
 	    if(arr->vectorAlleles->at(1).getAltCount() != 0 ){
 		if(arr->vectorAlleles->at(1).getRefCount() != 0 ){
-		    cerr<<"Cannot determine the root allele for "<<*arr<<endl;
+		    char b=arr->vectorAlleles->at(1).generateRandomAlleleBias(arr->ref,arr->alt);
+		    ancIsRef = (b==arr->ref);
+		    //cerr<<"Cannot determine the root allele for "<<*arr<<endl;
 		    //return 1;      
-		    continue;
+		    //continue;
 		}	
 		ancIsRef=false;
 	    }else{

diff --git a/ACF2BETASCAN.h b/ACF2BETASCAN.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/ACF2BPLINK.h b/ACF2BPLINK.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/ACF2EIGENSTRAT.cpp b/ACF2EIGENSTRAT.cpp
@@ -146,17 +146,22 @@ int ACF2EIGENSTRAT::run(int argc, char *argv[]){
 
 	unsigned int firstIndex=2;
 
-	if(printRoot)
-	    if(haploidRoot)
+	if(printRoot){
+	    if(haploidRoot){
 		genoFileS<<record->vectorAlleles->at(0).printEIGENSTRAT( false );	   
-	    else
-		genoFileS<<record->vectorAlleles->at(0).printEIGENSTRAT( !singleAlHomo );	   
-	if(printAnc)
-	    if(haploidRoot)
+	    }else{
+		genoFileS<<record->vectorAlleles->at(0).printEIGENSTRAT( !singleAlHomo );
+	    }
+	}
+
+	if(printAnc){
+	    if(haploidRoot){
 		genoFileS<<record->vectorAlleles->at(1).printEIGENSTRAT(false);	   
-	    else
+	    }else{
 		genoFileS<<record->vectorAlleles->at(1).printEIGENSTRAT( !singleAlHomo );
-
+	    }
+	}	
+
 	for(unsigned int i=firstIndex;i<record->vectorAlleles->size();i++){
 	    genoFileS<<record->vectorAlleles->at(i).printEIGENSTRAT( !singleAlHomo );	   
 	} 

diff --git a/ACF2EIGENSTRAT.h b/ACF2EIGENSTRAT.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/ACF2FASTA.h b/ACF2FASTA.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/ACF2GPHOCS.h b/ACF2GPHOCS.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/ACF2GROSS.h b/ACF2GROSS.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/ACF2NEXUS.h b/ACF2NEXUS.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/ACF2TREEMIX.h b/ACF2TREEMIX.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/AXT2ACF.h b/AXT2ACF.h
@@ -12,7 +12,7 @@
 #include <climits>
 #include <gzstream.h>
 
-#include "utils.h"
+#include "libgab.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"
 #include "AlleleRecords.h"

diff --git a/AvgCoaAlleleCounter.h b/AvgCoaAlleleCounter.h
@@ -13,7 +13,7 @@
 #include <math.h>
 #include <limits>
 
-#include "utils.h"
+#include "libgab.h"
 //#include <iostream>
 //#include <fstream>
 using namespace std;

diff --git a/AvgCoaResult.h b/AvgCoaResult.h
@@ -14,7 +14,7 @@ using namespace std;
 #include <vector>
 #include <list>
 #include "AvgCoaAlleleCounter.h"
-#include "utils.h"
+#include "libgab.h"
 
 class AvgCoaResult{
  private:

diff --git a/BAM2ACF.cpp b/BAM2ACF.cpp
@@ -187,6 +187,7 @@ int BAM2ACF::run(int argc, char *argv[]){
 
 	if( string(argv[i]) == "--bed"){
 	    bedfilename = string(argv[i+1]);
+	    i++;
 	    continue;
 	}
 
@@ -994,7 +995,7 @@ int BAM2ACF::run(int argc, char *argv[]){
     if (bed) bed_destroy(bed);
 
 
-    cerr<<"Program bam2acf terminated gracefully, considered "<<totalBasesL<<" bases at "<<totalSitesL<<endl;
+    cerr<<"Program bam2acf terminated gracefully, considered "<<totalBasesL<<" bases at "<<totalSitesL<<" sites"<<endl;
 
      //clean up
      //reader.Close();

diff --git a/BAM2ACF.h b/BAM2ACF.h
@@ -58,7 +58,7 @@ typedef struct {     // auxiliary data structure
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/BEAGLE2GLF.h b/BEAGLE2GLF.h
@@ -19,7 +19,7 @@
 /* #include <utils/bamtools_utilities.h> */
 
 
-#include "utils.h"
+#include "libgab.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"
 #include "AlleleRecords.h"

diff --git a/BPLINK2ACF.h b/BPLINK2ACF.h
@@ -19,7 +19,7 @@
 /* #include <utils/bamtools_utilities.h> */
 
 
-#include "utils.h"
+#include "libgab.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"
 #include "AlleleRecords.h"

diff --git a/CoreVCF.h b/CoreVCF.h
@@ -13,7 +13,7 @@
 #include <string.h>
 #include <vector> 
 #include <map>
-#include "utils.h"
+#include "libgab.h"
 
 using namespace std;
 

diff --git a/DistAlleleCounter.h b/DistAlleleCounter.h
@@ -15,7 +15,7 @@
 #include <vector>
 #include <sstream>
 
-#include "utils.h"
+#include "libgab.h"
 
 //#include <iostream>
 //#include <fstream>

diff --git a/DistResult.h b/DistResult.h
@@ -14,7 +14,7 @@ using namespace std;
 #include <vector>
 #include <list>
 #include "DistAlleleCounter.h"
-#include "utils.h"
+#include "libgab.h"
 
 class DistResult{
  private:

diff --git a/DstatCounter.h b/DstatCounter.h
@@ -15,7 +15,7 @@
 #include <sys/time.h>
 #include <stdlib.h>
 
-#include "utils.h"
+#include "libgab.h"
 
 //#include <iostream>
 //#include <fstream>

diff --git a/Dstat_core.h b/Dstat_core.h
@@ -9,7 +9,7 @@
 #define Dstat_core_h
 #include <ctype.h>
 #include "DstatResult.h"
-#include "utils.h"
+#include "libgab.h"
 
 using namespace std;
 

diff --git a/EIGENSTRAT2ACF.h b/EIGENSTRAT2ACF.h
@@ -19,7 +19,7 @@
 /* #include <utils/bamtools_utilities.h> */
 
 
-#include "utils.h"
+#include "libgab.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"
 #include "AlleleRecords.h"

diff --git a/F2Counter.h b/F2Counter.h
@@ -15,7 +15,7 @@
 #include <sys/time.h>
 #include <stdlib.h>
 
-#include "utils.h"
+#include "libgab.h"
 
 //#include <iostream>
 //#include <fstream>

diff --git a/F2_core.h b/F2_core.h
@@ -9,7 +9,7 @@
 #define F2_core_h
 #include <ctype.h>
 #include "F2Result.h"
-#include "utils.h"
+#include "libgab.h"
 
 using namespace std;
 

diff --git a/F3Counter.h b/F3Counter.h
@@ -15,7 +15,7 @@
 #include <sys/time.h>
 #include <stdlib.h>
 
-#include "utils.h"
+#include "libgab.h"
 
 //#include <iostream>
 //#include <fstream>

diff --git a/F3_core.h b/F3_core.h
@@ -9,7 +9,7 @@
 #define F3_core_h
 #include <ctype.h>
 #include "F3Result.h"
-#include "utils.h"
+#include "libgab.h"
 
 using namespace std;
 

diff --git a/FilterVCF.cpp b/FilterVCF.cpp
@@ -109,8 +109,10 @@ bool passedFilters(SimpleVCF * smvcf,const SetVCFFilters * filtersToUse){
 	coverageREF=smvcf->getDepthInfo();
     }
     //cerr<<"coverageREF\t"<<coverageREF<<endl;
-    if(coverageREF < filtersToUse->getMinCovcutoff() ||
-       coverageREF > filtersToUse->getMaxCovcutoff() ){
+    if( (coverageREF != -1)
+	&&
+	((coverageREF < filtersToUse->getMinCovcutoff()) ||
+	 (coverageREF > filtersToUse->getMaxCovcutoff() ))){
 	rejectLOWCOV_REF++;
 	return false;
     }

diff --git a/FstResult.h b/FstResult.h
@@ -14,7 +14,7 @@ using namespace std;
 #include <vector>
 #include <list>
 #include "FstAlleleCounter.h"
-#include "utils.h"
+#include "libgab.h"
 
 class FstResult{
  private:

diff --git a/GLF2ACF.h b/GLF2ACF.h
@@ -10,7 +10,7 @@
 
 #include <string>
 
-#include "utils.h"
+#include "libgab.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"
 #include "AlleleRecords.h"

diff --git a/GenomicRange.h b/GenomicRange.h
@@ -12,7 +12,7 @@
 #include <string>
 #include <stdlib.h>
 
-#include "utils.h"
+#include "libgab.h"
 
 using namespace std;
 

diff --git a/GenomicWindows.h b/GenomicWindows.h
@@ -15,7 +15,7 @@
 #include <fstream>
 #include <sys/time.h>
 
-#include "utils.h"
+#include "libgab.h"
 #include "GenomicRange.h"
 #include "RandomGenomicCoord.h"
 

diff --git a/Glac2BED.h b/Glac2BED.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"

diff --git a/Glac2FREQSPEC.cpp b/Glac2FREQSPEC.cpp
@@ -119,8 +119,10 @@ int Glac2FREQSPEC::run(int argc, char *argv[]){
 
 	    if(arr->vectorAlleles->at(0).getAltCount() != 0 ){
 		if(arr->vectorAlleles->at(0).getRefCount() != 0 ){
-		    cerr<<"Cannot determine the root allele for "<<*arr<<endl;
-		    return 1;      
+		    char b=arr->vectorAlleles->at(0).generateRandomAlleleBias(arr->ref,arr->alt);
+		    rootIsRef = (b==arr->ref);
+		    // cerr<<"Cannot determine the root allele for "<<*arr<<endl;
+		    // return 1;      
 		}	
 		rootIsRef=false;
 	    }else{
@@ -137,8 +139,10 @@ int Glac2FREQSPEC::run(int argc, char *argv[]){
 
 	    if(arr->vectorAlleles->at(1).getAltCount() != 0 ){
 		if(arr->vectorAlleles->at(1).getRefCount() != 0 ){
-		    cerr<<"Cannot determine the root allele for "<<*arr<<endl;
-		    return 1;      
+		    char b=arr->vectorAlleles->at(1).generateRandomAlleleBias(arr->ref,arr->alt);
+		    ancIsRef = (b==arr->ref);
+		    // cerr<<"Cannot determine the root allele for "<<*arr<<endl;
+		    // return 1;      
 		}	
 		ancIsRef=false;
 	    }else{

diff --git a/Glac2FREQSPEC.h b/Glac2FREQSPEC.h
@@ -16,7 +16,7 @@
 #include <string>
 #include <vector>
 
-#include "utils.h"
+#include "libgab.h"
 #include "ReadTabix.h"
 #include "GlacWriter.h"
 #include "GlacParser.h"