def loadData(filename="",sz=100,useRand=False): if len(filename)==0: filename="../../GWAS/cleaned"; [y,sFil]=getData(filename); if useRand: rand.shuffle(y); I=[]; I1=[i for i in range(0,len(y)) if y[i]==1]; I2=[i for i in range(0,len(y)) if y[i]!=1]; I.extend(I1[:sz]); I.extend(I2[:sz]); NI=[]; NI.extend(I1[sz:]) NI.extend(I2[sz:]) y1=[y[i] for i in I] sFil1=sFil[I,:]; y2=[y[i] for i in NI] sFil2=sFil[NI,:]; return [[y1,sFil1],[y2,sFil2]];
def NoisySig(mret,eps,filename,savename=""):
if len(savename)==0:
savename="Output/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 EIGN!"
EIGN=EIGN_STRAT(BED,2);
sc=EIGN.prod(y);
sc=[abs(s) for s in sc];
sc=sorted(sc,reverse=True);
print "get True!";
truSNPs=pt(y,EIGN,mret,-1,algor="noise");
our=[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)];
neighbor=[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;
nosySNPs=pt(y,EIGN,mret,e,algor="DPE",reuse=True);
our[i]=our[i]+inter(truSNPs,noisySNPs)/float(reps);
noisySNPs=pt(y,MU,mret,e,algor="score");
score[i]=score[i]+inter(truSNPs,noisySNPs)/float(reps);
noisySNPs=pt(y,MU,mret,e,algor="noise");
noise[i]=noise[i]+inter(truSNPs,noisySNPs)/float(reps);
noisySNPs=pt(y,MU,mret,e,algor="neighbor");
neighbor=neighbor[i]+inter(truSNPs,noisySNPs)/float(reps);
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();
def plotWald(eps,filename,savename="",k=5): if len(savename)==0: savename="OutputDir/res_wald_"+str(eps[0])+"_"+str(k)+".txt" print "load Data" [y,BED]=ld.getData(filename); print "calc MU!" MU=MU_STRAT(BED,k); print "get True!"; tru=wt(y,MU,-1,snps=[],forFigs=False); fil=open(savename,"w"); n=len(y); for i in range(0,10): e=eps[i]; print e; res=wt(y,MU,e,snps=[],forFigs=True); err=sorted([float(n-k-1)*abs(res[i]-tru[i]) for i in range(0,len(tru))]); m=len(err); med=err[int(.5*m)] up=err[int(.75*m)]; down=err[int(.25*m)]; print med; fil.write(str(e)+" "+str(down)+" "+str(med)+" "+str(up)+"\n"); fil.close();
print "As it stands: "
print "BedFile: " + bedFil
print "epsilon: " + str(epsilon)
print "Type: " + typ
if typ != "Count":
print "Aglorithm: " + algor
if typ == "Top":
print "mret: " + str(mret)
if typ == "Wald":
print "SNPs: " + str(snps)
if typ == "Count":
print "pvals: " + str(pval)
print "\n\n\n"
print "Load Data!"
[y, BED] = getData(bedFil)
print "Calculating MU matrix"
if num < 1:
MU = MU_LMM(BED, (se2, sg2))
else:
MU = MU_LMM(BED, [num, epsilon])
if typ == "Top":
PrivGWAS.Top(MU, y, epsilon, mret, algor, savename)
elif typ == "Count":
PrivGWAS.count(MU, y, epsilon, pval, savename)
elif typ == "Wald":
PrivGWAS.wald(MU, y, epsilon, snps, savename)
elif typ == "Herit":
print "The estimated heritability:"
print "Sigma_e^2 is " + str(MU.se2)
print "BedFile: "+bedFil;
print "epsilon: "+str(epsilon);
print "Type: "+typ;
if typ!="Count":
print "Aglorithm: "+algor;
if typ=="Top":
print "mret: "+str(mret);
if typ=="Wald":
print "SNPs: "+str(snps);
if typ=="Count":
print "threshold: "+str(pval);
if exact:
print "Use exact method!"
print "\n\n\n";
print "Load Data!"
[y,BED]=getData(bedFil);
print "Calculating MU matrix"
MU=MU_STRAT(BED,k);
if exact:
MU.calcMU(k,exact=True);
n=len(y);
if typ=="Top":
PrivGWAS.Top(MU,y,epsilon,mret,algor,savename,snpList);
elif typ=="Count":
PrivGWAS.count(MU,y,epsilon,pval,savename);
elif typ=="Wald":
PrivGWAS.wald(MU,y,epsilon,snps,savename,coeff=float(n-k-1));
print "Type: " + typ
if typ != "Count":
print "Aglorithm: " + algor
if typ == "Top":
print "mret: " + str(mret)
if typ == "Wald":
print "SNPs: " + str(snps)
if typ == "Count":
print "pvals: " + str(pval)
print "\n\n\n"
print "Load Data!"
y = []
X = []
sFil = []
Q = []
[y, X, Q, sFil] = getData(bedFil, useCov)
print "Calculating MU matrix"
MU = DP.getMU(y, X, Q=Q, se2=se2, sg2=sg2, k=k, meth=meth)
if typ == "Top":
picks = DP.PickTopSNP(y,
X,
mret,
epsilon=epsilon,
k=k,
se2=se2,
sg2=sg2,
MU=MU,
meth=meth,
algor=algor)
if len(picks) == 0:
print "Bad argument!"
print "Method: "+meth; print "BedFile: "+bedFil; print "epsilon: "+str(epsilon); print "Type: "+typ; if typ!="Count": print "Aglorithm: "+algor; if typ=="Top": print "mret: "+str(mret); if typ=="Wald": print "SNPs: "+str(snps); if typ=="Count": print "pvals: "+str(pval); print "\n\n\n"; print "Load Data!" y=[];X=[];sFil=[];Q=[]; [y,X,Q,sFil]=getData(bedFil,useCov); print "Calculating MU matrix" MU=DP.getMU(y,X,Q=Q,se2=se2,sg2=sg2,k=k,meth=meth); if typ=="Top": picks=DP.PickTopSNP(y,X,mret,epsilon=epsilon,k=k,se2=se2,sg2=sg2,MU=MU,meth=meth,algor=algor) if len(picks)==0: print "Bad argument!" return; print "The mret top scoring SNPs:" for i in picks: print sFil.sid[i]; print sFil.pos[i] elif typ=="Wald": picks=sFil.sid_to_index(snps);#[snps.index(i) for i in snps]; Scores=DP.estWald(y,X,epsilon,k=k,snps=picks,MU=MU,meth=meth,algor=algor); print "The estimated Wald scores are:";
