Merge pull request #389 from ANGSD/mcall

Merging the multiallelic caller. This is for internal testing and comparison and is not finalized yet.

abc.cpp

@@ -34,6 +34,7 @@
 #include "abcPSMC.h"
 #include "abcScounts.h"
 #include "abcWriteBcf.h"
+#include "abcMcall.h"
 //below we set some variables that are shared between all the analysis classes
 #define MAX_CLASS 35
 
@@ -54,33 +55,34 @@ abc **extra(int &nItem,const char *outfiles,int inputtype,argStruct *arguments){
   tskStuff[nit++] = new abcCounts(outfiles,arguments,inputtype);//2
   tskStuff[nit++] = new abcError(outfiles,arguments,inputtype);//3
   tskStuff[nit++] = new abcGL(outfiles,arguments,inputtype);//4
-  tskStuff[nit++] = new abcMajorMinor(outfiles,arguments,inputtype);//5
-  tskStuff[nit++] = new abcFreq(outfiles,arguments,inputtype);//6
-  tskStuff[nit++] = new abcAsso(outfiles,arguments,inputtype);//7
-  tskStuff[nit++] = new abcHWE(outfiles,arguments,inputtype); // 8
-  tskStuff[nit++] = new abcAncError(outfiles,arguments,inputtype);//9
-  tskStuff[nit++] = new abcCallGenotypes(outfiles,arguments,inputtype);//10
+  tskStuff[nit++] = new abcMcall(outfiles,arguments,inputtype);//5
+  tskStuff[nit++] = new abcMajorMinor(outfiles,arguments,inputtype);//6
+  tskStuff[nit++] = new abcFreq(outfiles,arguments,inputtype);//7
+  tskStuff[nit++] = new abcAsso(outfiles,arguments,inputtype);//8
+  tskStuff[nit++] = new abcHWE(outfiles,arguments,inputtype); // 9
+  tskStuff[nit++] = new abcAncError(outfiles,arguments,inputtype);//10
+  tskStuff[nit++] = new abcCallGenotypes(outfiles,arguments,inputtype);//11
   tskStuff[nit++] = new abcSaf(outfiles,arguments,inputtype);
   tskStuff[nit++] = new abcCovar(outfiles,arguments);
   tskStuff[nit++] = new abcTsk(outfiles,arguments,inputtype);
-  tskStuff[nit++] = new abcFilterSNP(outfiles,arguments,inputtype);//14
+  tskStuff[nit++] = new abcFilterSNP(outfiles,arguments,inputtype);//15
   tskStuff[nit++] = new abcSnpTools(outfiles,arguments,inputtype);
   tskStuff[nit++] = new abcHetPlas(outfiles,arguments,inputtype);
   tskStuff[nit++] = new abcWritePlink(outfiles,arguments,inputtype);
-  tskStuff[nit++] = new abcDstat(outfiles,arguments,inputtype);//18
+  tskStuff[nit++] = new abcDstat(outfiles,arguments,inputtype);//19
   tskStuff[nit++] = new abcWriteFasta(outfiles,arguments,inputtype);
   tskStuff[nit++] = new abcSmartCounts(outfiles,arguments,inputtype);
   tskStuff[nit++] = new abcTemplate(outfiles,arguments,inputtype);
-  tskStuff[nit++] = new abcWriteVcf(outfiles,arguments,inputtype);//22
+  tskStuff[nit++] = new abcWriteVcf(outfiles,arguments,inputtype);//23
   tskStuff[nit++] = new abcHaploCall(outfiles,arguments,inputtype);
   tskStuff[nit++] = new abcDstat2(outfiles,arguments,inputtype);
-  tskStuff[nit++] = new abcHWE_F(outfiles,arguments,inputtype); // 25
+  tskStuff[nit++] = new abcHWE_F(outfiles,arguments,inputtype); // 26
   tskStuff[nit++] = new abcIBS(outfiles,arguments,inputtype); // 
   tskStuff[nit++] = new abcPSMC(outfiles,arguments,inputtype); //
   tskStuff[nit++] = new abcScounts(outfiles,arguments,inputtype); //
-  tskStuff[nit++] = new abcWriteBcf(outfiles,arguments,inputtype); // 29
-  tskStuff[nit++] = new abcRAD(outfiles,arguments,inputtype); // 30
-
+  tskStuff[nit++] = new abcWriteBcf(outfiles,arguments,inputtype); // 30
+  tskStuff[nit++] = new abcRAD(outfiles,arguments,inputtype); // 31
+  //remember to update changeChr in shared.cpp if order gets changed
   nItem = nit;
   return tskStuff;
 }

abcAsso.cpp

@@ -739,7 +739,7 @@ void abcAsso::scoreAsso(funkyPars  *pars,assoStruct *assoc){
       for(int i=0 ; i<pars->nInd ;i++)
 	y[i]=ymat.matrix[i][yi]; 
 
-      freqStruct *freq = (freqStruct *) pars->extras[6];
+      freqStruct *freq = (freqStruct *) pars->extras[7];
       stat[yi][s]=doAssociation(pars,pars->post[s],y,keepInd[yi][s],keepList,freq->freq[s],s,assoc);
 
       //cleanup
@@ -1239,7 +1239,7 @@ void abcAsso::dosageAsso(funkyPars  *pars,assoStruct *assoc){
       for(int i=0 ; i<pars->nInd ;i++)
 	y[i]=ymat.matrix[i][yi]; 
 
-      freqStruct *freq = (freqStruct *) pars->extras[6];
+      freqStruct *freq = (freqStruct *) pars->extras[7];
 
       stat[yi][s]=dosageAssoc(pars,&design,&designNull,pars->post[s],y,keepInd[yi][s],keepList,freq->freq[s],s,assoc,model,isBinary,isCount,start,1);
 
@@ -2259,7 +2259,7 @@ void abcAsso::emAsso(funkyPars  *pars,assoStruct *assoc){
       for(int i=0 ; i<pars->nInd ;i++)
 	y[i]=ymat.matrix[i][yi]; 
 
-      freqStruct *freq = (freqStruct *) pars->extras[6];
+      freqStruct *freq = (freqStruct *) pars->extras[7];
 
       stat[yi][s]=doEMasso(pars,&design,&designNull,&postAll,pars->post[s],y,keepInd[yi][s],keepList,freq->freq[s],s,assoc,model,isBinary,isCount,start,1);
 
@@ -2560,7 +2560,7 @@ void abcAsso::emAssoWald(funkyPars  *pars,assoStruct *assoc){
       for(int i=0 ; i<pars->nInd ;i++)
 	y[i]=ymat.matrix[i][yi]; 
 
-      freqStruct *freq = (freqStruct *) pars->extras[6];
+      freqStruct *freq = (freqStruct *) pars->extras[7];
 
       tmp=doEMassoWald(pars,&design,&postAll,pars->post[s],y,keepInd[yi][s],keepList,freq->freq[s],s,assoc,model,isBinary,isCount,start,1);            
       //if not enough, WT, HE or HO or ind to run test
@@ -2675,7 +2675,7 @@ void abcAsso::hybridAsso(funkyPars  *pars,assoStruct *assoc){
       for(int i=0 ; i<pars->nInd ;i++)
 	y[i]=ymat.matrix[i][yi]; 
 
-      freqStruct *freq = (freqStruct *) pars->extras[6];
+      freqStruct *freq = (freqStruct *) pars->extras[7];
       stat[yi][s]=doAssociation(pars,pars->post[s],y,keepInd[yi][s],keepList,freq->freq[s],s,assoc);
 
       // cutoff has been changed to p-value
@@ -3589,7 +3589,7 @@ void abcAsso::printDoAsso(funkyPars *pars){
   if(doPrint)
     fprintf(stderr,"staring [%s]\t[%s]\n",__FILE__,__FUNCTION__);
 
-  freqStruct *freq = (freqStruct *) pars->extras[6];
+  freqStruct *freq = (freqStruct *) pars->extras[7];
   assoStruct *assoc= (assoStruct *) pars->extras[index];
 
   //chisq distribution with df=1

abcCovar.cpp

@@ -137,7 +137,7 @@ void abcCovar::run(funkyPars *pars){
   for(int s=0;s<pars->numSites;s++)
     Egeno[s] = new double[pars->nInd];
 
-  freqStruct *freq = (freqStruct *) pars->extras[6];
+  freqStruct *freq = (freqStruct *) pars->extras[7];
   //maybe change to posterior based on frequency
   for(int s=0;s<pars->numSites;s++){
     keepSNP[s]=0;

abcFilterSNP.cpp

@@ -232,7 +232,7 @@ void abcFilterSNP::run(funkyPars *pars){
       if(sb_pval!=-1 && sb3(cnts)<sb_pval)
 	pars->keepSites[s] = 0;
 
-      funkyHWE *hweStruct = (funkyHWE *) pars->extras[8];//THIS IS VERY NASTY! the ordering within general.cpp is now important
+      funkyHWE *hweStruct = (funkyHWE *) pars->extras[9];//THIS IS VERY NASTY! the ordering within general.cpp is now important
       double lrt = 2*hweStruct->like0[s]-2*hweStruct->likeF[s];
       double pval;
       if(std::isnan(lrt))
@@ -263,7 +263,7 @@ void abcFilterSNP::run(funkyPars *pars){
       if(2*phi(Z)<edge_pval)
 	pars->keepSites[s] = 0;
 
-      genoCalls *gcw =(genoCalls *) pars->extras[10];
+      genoCalls *gcw =(genoCalls *) pars->extras[11];
       int **gc=NULL;
       if(gcw)
 	gc = gcw->dat;

abcGL.cpp

@@ -377,7 +377,7 @@ void abcGL::run(funkyPars *pars){
   if(soap.doRecal!=1 || ancestral_lik.doRecal!=1)
     likes = new double*[pars->chk->nSites];
   if(GL==1)
-    call_bam(pars->chk,likes,trim);
+    call_bam(pars->chk,likes,trim,pars->keepSites);
   else if(GL==2)
     call_gatk(pars->chk,likes,trim);
   else if(GL==3){
@@ -497,7 +497,8 @@ void abcGL::printLike(funkyPars *pars) {
     //beagle format
     bufstr.l = 0; //set tmpbuf beginning to zero
     for(int s=0;s<pars->numSites;s++) {
-      lh3struct *lh3 = (lh3struct*) pars->extras[index+1];
+      lh3struct *lh3 = (lh3struct*) pars->extras[index+2];
+      assert(lh3);
       if(pars->keepSites[s]==0||lh3->hasAlloced[s]==0)
 	continue;
 

abcHWE_F.cpp

@@ -117,7 +117,7 @@ void abcHWE_F::print(funkyPars *pars){
   for(int s=0;s<pars->numSites;s++){
     if(pars->keepSites[s]==0) 
       continue;
-    freqStruct *freq = (freqStruct *) pars->extras[6];
+    freqStruct *freq = (freqStruct *) pars->extras[7];
     float lrt= 2*hweStruct->like0[s]-2*hweStruct->likeF[s];
     //DRAGON lrt is sometimes nan
     float pval;
@@ -147,7 +147,7 @@ void abcHWE_F::run(funkyPars *pars){
   double *F = new double[pars->numSites];
   double *like0 = new double[pars->numSites];
   double *likeF = new double[pars->numSites];
-  freqStruct *freqS = (freqStruct *) pars->extras[6];
+  freqStruct *freqS = (freqStruct *) pars->extras[7];
 
   double **loglike3;
   loglike3=angsd::get3likesRescale(pars);