-
Notifications
You must be signed in to change notification settings - Fork 1
/
simulations_JAGUAR_vs_JT.R
143 lines (128 loc) · 5.06 KB
/
simulations_JAGUAR_vs_JT.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
#!/usr/bin/env Rscript
###########################################################################################
## Run simulations testing our joint model against JAGUAR
## both in the presence and absence of methylation data.
##
## Author: Chaitanya Acharya
## Updated: March 7, 2016
###########################################################################################
## Load the following libraries
lib.list=c("lme4","plyr","mvtnorm","Rcpp","RcppArmadillo","compiler");
for(i in 1:length(lib.list)){
if(any(installed.packages()[,1]==lib.list[i])){
library(lib.list[i],character.only=T)}else{
source("http://bioconductor.org/biocLite.R");
biocLite(lib.list[i]);
library(lib.list[i],character.only=T)};
}
sourceCpp("joint_test.cpp")
sourceCpp("jaguar.cpp")
## main function
main = function(nobs=100,k=5,maf=0.10,mu=5,lambda=0,tau=1,theta=0,eps=1,
bta=0,gamma=0,delta=0,phi=0){
snp = rbinom(nobs,2,maf);
v = rnorm(k,0,gamma);
w = rnorm(k,0,delta);
x = rnorm(k,0,theta);
alpha = rep(mu,k);
omega = matrix(rnorm(k*k),k)
meth_dat = rmvnorm(nobs,rep(0,k),as.matrix(nearPD(forceSymmetric(omega))$mat))
Y = do.call("rbind",lapply(1:nobs,function(i){
alpha+bta*snp[i] + lambda*meth_dat[i,] + phi*snp[i]*meth_dat[i,] + rnorm(1,0,tau) + v*snp[i] + w*snp[i]*meth_dat[i,] + x*meth_dat[i,] + rnorm(k,0,eps) }
));
data = data.frame("IND"=as.factor(rep(1:nobs,each=k)),"Gene"=as.vector(t(Y)),"Geno" = rep(snp,each=k),"Tissue"=as.factor(rep(1:k,nobs)),"Meth"=as.vector(t(meth_dat)));
fit = suppressWarnings(lmer(Gene~0+Tissue+Meth+(1|IND)+(0+Meth|Tissue),data,REML=F));
est_eps = sigma(fit)^2; est_tau = VarCorr(fit)[[1]][1]; est_theta = VarCorr(fit)[[2]][1];
Yhat = data$Gene - model.matrix(fit) %*% fixef(fit);
A = as.matrix(getME(fit,"Z"))[,1:nobs];
Z = matrix(rep(snp-mean(snp),each=k),nobs*k,1);
geno.list = meth.list = meth.geno.list = list();
for(i in 1:nobs){
geno.list[[i]] = (snp[i]-mean(snp))*diag(k);
meth.list[[i]] = meth_dat[i,]*diag(k);
meth.geno.list[[i]] = ( (snp[i]-mean(snp))*meth_dat[i,] )*diag(k);
}
MG = matrix( (Z*data$Meth)-mean(Z*data$Meth),nobs*k,1);
B = do.call("rbind",geno.list);
C = do.call("rbind",meth.geno.list);
D = do.call("rbind",meth.list);
V = as.matrix( (est_eps * diag(nobs*k)) + (est_tau * tcrossprod(A)) + (est_theta * tcrossprod(D)) );
JT = joint_test2(as.matrix(Yhat),Z,MG,B,C,V);
fit = suppressWarnings(lmer(Gene~0+Tissue+(1|IND),data,REML=F))
est.eps = sigma(fit)^2; est.tau = VarCorr(fit)[[1]][1]
Ynew = apply(Y,1,function(x) x-colMeans(Y))
U.pval = jagSIM(est.eps,est.tau,k,Ynew,snp)
subData = split(data,data$Tissue);
linear_eQTL = min(do.call("cbind",lapply(1:length(subData),function(i) coef(summary(lm(Gene~Geno,subData[[i]])))[2,4] )));
return(c("TBT_noMeth"=linear_eQTL,"JAGUAR"=U.pval,"U_psi"=JT))
}
nsim=1000
test = cmpfun(main)
##
# When methylation is not important or methylation data is not available
##
nt = 5
maf = 0.30
gamma = c(0,0.15,0.25,0.5)
bta = c(0,0.5)
outGAM = vector(mode="list",length=length(gamma))
outBTA = vector(mode="list",length=length(bta))
for(b in 1:length(bta)){
cat("\n BTA: ",bta[b],"\t")
for(g in 1:length(gamma)){
cat("GAMMA: ",gamma[g],"\t")
FUN = do.call("rbind",rlply(nsim,test(k=nt,bta=bta[b],gamma=gamma[g])))
power =c("TBT"=sum(FUN[,1]<=(0.05/nt))/nsim, apply(FUN[,-1],2,function(x) sum(x<=0.05)/nsim))
outGAM[[g]] = c("MAF"=maf,"BTA"=bta[b],power)
}
outBTA[[b]] = do.call("rbind",outGAM)
cat("\n")
}
out=do.call("rbind",outBTA)
out
##
# When methylation is important or in the presence of methylation data
##
nt = c(5)
maf= c(0.30)
bta = phi = c(0,0.25)
gamma = delta = c(0,0.15,0.25)
outMAF = vector(mode="list",length=length(maf))
outGAM = vector(mode="list",length=length(gamma))
outDELTA = vector(mode="list",length=length(delta))
outBTA = vector(mode="list",length=length(bta))
outPHI = vector(mode="list",length=length(phi))
out = vector(mode="list",length=length(nt))
for(i in 1:length(nt)){
cat("Number of tissues: ",nt[i],"\n")
for(j in 1:length(maf)){
cat("\n MAF: ",maf[j],"\t")
for(b in 1:length(bta)){
cat("\n BTA: ",bta[b],"\t")
for(d in 1:length(delta)){
cat("\n DELTA: ",delta[d],"\t")
for(g in 1:length(gamma)){
cat("\n GAM: ",gamma[g],"\t")
for(p in 1:length(phi)){
cat("PHI: ",phi[p],"\t")
FUN = do.call("rbind",rlply(nsim,test(nobs=100,lambda=0.5,theta=0.5,k=nt[i],bta=bta[b],gamma=gamma[g],maf=maf[j],delta=delta[d],phi=phi[p])))
power =c("TBT"=sum(FUN[,1]<=(0.05/nt))/nsim, apply(FUN[,-1],2,function(x) sum(x<=0.05)/nsim))
testOUT = c("PHI"=phi[p],power)
outPHI[[p]] = c("MAF"=maf[j],"BTA"=bta[b],"DELTA"=delta[d],"GAM"=gamma[g],testOUT)
}
outGAM[[g]] = do.call("rbind",outPHI)
}
outDELTA[[d]] = do.call("rbind",outGAM)
}
outBTA[[b]] = do.call("rbind",outDELTA)
}
outMAF[[j]] = do.call("rbind",outBTA)
}
out[[i]] = do.call("rbind",outMAF)
names(out)[[i]] = nt[i]
colnames(out[[i]])[6:8]=c("TBT","JAG","Upsi")
filename = paste("JAG_vs_JOINT_",nt[i],"_",format(Sys.Date(),format="%b%d.%Y"),".txt",sep="")
write.table(out[[i]],filename,sep="\t",row.names=F,quote=F)
cat("\n")
}
out