def RunGWAS(filename,k=5,eps=2.0):
sz=450
[[y1,sFil1],[y2,sFil2]]=loadData(filename=filename,useRand=False,sz=sz);
MU1=MU_STRAT(sFil1,k);
MU2=MU_STRAT(sFil2,k);
print "EXACT1!"
MU1.calcMU(k,exact=True)
print "Exact 2!"
MU2.calcMU(k,exact=True);
res=pt(y1,MU1,1,-1,algor="noise");
resL=pt(y1,MU1,20,-1,algor="noise");
print resL
#print wt(y1,MU1,-1,snps=res);
res2=pt(y2,MU2,1,-1,algor="noise");
print res;
resT=res;
print res2;
resT=wt(y2,MU2,eps,snps=resL[5:7]);
res2=wt(y1,MU1,eps,snps=res);
res3=wt(y1,MU1,-1,snps=[]);
res4=wt(y2,MU2,-1,snps=[]);
n=len(y2);
print len(y1);
print n;
val=[resT[0]*(n-k-1),resT[1]*(n-k-1)]
n2=len(y1);
print "In Val:"
print val;
print "In Orig"
print res2[0]*(n2-k-1);
n2=len(y1);
print np.median(res3)*(n2-k-1)
n2=len(y1);
res3=sorted(res3,reverse=True)[:10];
res3=[(n2-k-1)*i for i in res3];
print res3
fil=open("saveRes.txt","w")
for r in res4:
fil.write(str(r)+"\n")
fil.close();
res4=sorted(res4,reverse=True)[:20];
res3=[(n-k-1)*i for i in res4];
print res3
CIlst=CI(y2,MU2,.95,2.0,resL[5:6]);
CIlst2=CI(y2,MU2,.95,2.0,resL[6:7]);
CIlst=[[(n-k-1)*i for i in s] for s in CIlst]
CIlst22=[[(n-k-1)*i for i in s] for s in CIlst2]
print CIlst;
print CIlst2;
def TestCase(cutoff=23,num=2,filename="../GWAS/NoFam_0.05",k=10,sz=450): [[y1,sFil1],[y2,sFil2]]=loadData(filename=filename,useRand=False,sz=sz); MU1=MU_STRAT(sFil1,k) MU2=MU_STRAT(sFil2,k) n=len(y2); n2=len(y1); print "Exact 1!" MU1.calcMU(k,exact=True); print "Exact 2!" MU2.calcMU(k,exact=True); print "Get top in set 1" top1=pt(y1,MU1,20,-1,algor="noise"); print "Get Scores" scores=wt(y1,MU1,-1,snps=top1); print "Get "+str(num)+" SNPs directly below "+str(cutoff); i=min([i for i in range(0,len(top1)) if scores[i]<cutoff/float(n2-k-1)]) print top1[i]; print float((n2-k-1)*scores[i]) #print i; #print scores[:10]; print ""; print top1[i+1]; print float((n2-k-1)*scores[i+1]) CIlst=CI(y2,MU2,.95,2.0,top1[i:i+2]); Clst=[[(n-k-1)*i for i in s] for s in CIlst] print Clst;
def plotTop(mret,eps,filename,savename=""):
if len(savename)==0:
savename="OutputDir/LMM_top_"+str(eps)+"_"+str(mret)+".txt"
epsilons=[eps*i for i in range(1,11)];
print "load Data"
[y,BED]=ld.getData(filename);
print "calc MU!"
MU=MU_LMM(BED,[1,-1.0]);
sc=MU.prod(y);
sc=[abs(s) for s in sc];
sc=sorted(sc,reverse=True);
print "get True!";
tru=pt(y,MU,mret,-1,algor="noise");
neighs=[0.0 for i in range(1,11)];
score=[0.0 for i in range(1,11)];
noise=[0.0 for i in range(1,11)];
reps=20;
for i in range(0,10):
e=epsilons[i];
print e;
for j in range(0,reps):
print j;
e1=e*mret/float(mret+1);
MU=MU_LMM(BED,[10,e/float(mret+1)]);
gs=pt(y,MU,mret,e1,algor="neighbor",reuse=True);
neighs[i]=neighs[i]+inter(tru,gs)/float(reps);
gs=pt(y,MU,mret,e1,algor="score");
score[i]=score[i]+inter(tru,gs)/float(reps);
gs=pt(y,MU,mret,e1,algor="noise");
noise[i]=noise[i]+inter(tru,gs)/float(reps);
fil=open(savename,"w");
fil.write("Testing Top SNPs with DP, "+filename);
fil.write("\nEpsilon:")
for i in range(0,10):
fil.write(" "+str(epsilons[i]))
fil.write("\nNoise:")
for i in range(0,10):
fil.write(" "+str(noise[i]))
fil.write("\nScore:")
for i in range(0,10):
fil.write(" "+str(score[i]))
fil.write("\nNeighbor:")
for i in range(0,10):
fil.write(" "+str(neighs[i]))
fil.close();
def plotTop(mret, eps, filename, savename=""):
if len(savename) == 0:
savename = "OutputDir/LMM_top_" + str(eps) + "_" + str(mret) + ".txt"
epsilons = [eps * i for i in range(1, 11)]
print "load Data"
[y, BED] = ld.getData(filename)
print "calc MU!"
MU = MU_LMM(BED, [1, -1.0])
sc = MU.prod(y)
sc = [abs(s) for s in sc]
sc = sorted(sc, reverse=True)
print "get True!"
tru = pt(y, MU, mret, -1, algor="noise")
neighs = [0.0 for i in range(1, 11)]
score = [0.0 for i in range(1, 11)]
noise = [0.0 for i in range(1, 11)]
reps = 20
for i in range(0, 10):
e = epsilons[i]
print e
for j in range(0, reps):
print j
e1 = e * mret / float(mret + 1)
MU = MU_LMM(BED, [10, e / float(mret + 1)])
gs = pt(y, MU, mret, e1, algor="neighbor", reuse=True)
neighs[i] = neighs[i] + inter(tru, gs) / float(reps)
gs = pt(y, MU, mret, e1, algor="score")
score[i] = score[i] + inter(tru, gs) / float(reps)
gs = pt(y, MU, mret, e1, algor="noise")
noise[i] = noise[i] + inter(tru, gs) / float(reps)
fil = open(savename, "w")
fil.write("Testing Top SNPs with DP, " + filename)
fil.write("\nEpsilon:")
for i in range(0, 10):
fil.write(" " + str(epsilons[i]))
fil.write("\nNoise:")
for i in range(0, 10):
fil.write(" " + str(noise[i]))
fil.write("\nScore:")
for i in range(0, 10):
fil.write(" " + str(score[i]))
fil.write("\nNeighbor:")
for i in range(0, 10):
fil.write(" " + str(neighs[i]))
fil.close()
def plotTop(mret,eps,filename,savename=""):
if len(savename)==0:
savename="OutputDir/res_top_"+str(eps)+"_"+str(mret)+".txt"
epsilons=[eps*i for i in range(1,11)];
print "load Data"
[y,BED]=ld.getData(filename);
print "calc MU!"
MU=MU_STRAT(BED,10);
sc=MU.prod(y);
sc=[abs(s) for s in sc];
sc=sorted(sc,reverse=True);
print "get True!";
tru=pt(y,MU,mret,-1,algor="noise");
neighs=[0.0 for i in range(1,11)];
score=[0.0 for i in range(1,11)];
noise=[0.0 for i in range(1,11)];
reps=20;
for i in range(0,10):
e=epsilons[i];
print e;
for j in range(0,reps):
print j;
gs=pt(y,MU,mret,e,algor="neighbor",reuse=True);
neighs[i]=neighs[i]+inter(tru,gs)/float(reps);
gs=pt(y,MU,mret,e,algor="score");
score[i]=score[i]+inter(tru,gs)/float(reps);
gs=pt(y,MU,mret,e,algor="noise");
noise[i]=noise[i]+inter(tru,gs)/float(reps);
fil=open(savename,"w");
fil.write("Testing Top SNPs with DP, "+filename);
fil.write("\nEpsilon:")
for i in range(0,10):
fil.write(" "+str(epsilons[i]))
fil.write("\nNoise:")
for i in range(0,10):
fil.write(" "+str(noise[i]))
fil.write("\nScore:")
for i in range(0,10):
fil.write(" "+str(score[i]))
fil.write("\nNeighbor:")
for i in range(0,10):
fil.write(" "+str(neighs[i]))
fil.close();
