| Title: | Multi-Modal Matrix Joint Factorization for Integrative Multi-Omics Data Analysis |
|---|---|
| Description: | Multi modality data matrices are factorized conjointly into the multiplication of a shared sub-matrix and multiple modality specific sub-matrices, group sparse constraint is applied to the shared sub-matrix to capture the homogeneous and heterogeneous information, respectively. Then the samples are classified by clustering the shared sub-matrix with kmeanspp(), a new version of kmeans() developed here to obtain concordant results. The package also provides the cluster number estimation by rotation cost. Moreover, cluster specific features could be retrieved using hypergeometric tests. |
| Authors: | Xiaoyao Yin [aut, cre] |
| Maintainer: | Xiaoyao Yin <[email protected]> |
| License: | GPL-3 |
| Version: | 0.1.0 |
| Built: | 2024-12-25 06:31:29 UTC |
| Source: | CRAN |
Calculate the cost defined by the objective function
cost(WL, init_list, lambda)cost(WL, init_list, lambda)
WL |
a list of multiple modality data matrices |
init_list |
a list of the initialized modality specific sub-matrices list Hi and shared sub-matrix E |
lambda |
a parameter to set the relative weight of the L1,infinity norm defined on sub-matrices list E |
res, a real data value of the cost
library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4) update_H_list <- update_H(temp_data,init_list) lambda <- 0.01 update_E_list <- update_E(temp_data,update_H_list,lambda) new_cost <- cost(temp_data,update_E_list,lambda)library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4) update_H_list <- update_H(temp_data,init_list) lambda <- 0.01 update_E_list <- update_E(temp_data,update_H_list,lambda) new_cost <- cost(temp_data,update_E_list,lambda)
Generate the simulated dataset with three modalities with the package crimmix
crimmix_data_gen(nclust=4, n_byClust=c(10,20,5,25), feature_nums=c(1000,500,5000), noises=c(0.5,0.01,0.3),props=c(0.005,0.01,0.02))crimmix_data_gen(nclust=4, n_byClust=c(10,20,5,25), feature_nums=c(1000,500,5000), noises=c(0.5,0.01,0.3),props=c(0.005,0.01,0.02))
nclust |
number of clusters |
n_byClust |
number of samples per cluster |
feature_nums |
number of features in each modality |
noises |
percentage of noise adding to each modality |
props |
proportion of cluster related features in each modality |
res, a list of length 2, where the first element is a list of simulated data, while the second element is a vector indicating the true label of each sample
crimmix_data <- crimmix_data_gen(nclust=4, n_byClust=c(10,20,5,25), feature_nums=c(1000,500,5000), noises=c(0.5,0.01,0.3),props=c(0.005,0.01,0.02))crimmix_data <- crimmix_data_gen(nclust=4, n_byClust=c(10,20,5,25), feature_nums=c(1000,500,5000), noises=c(0.5,0.01,0.3),props=c(0.005,0.01,0.02))
Screen the cluster related features via hypergeometric test p value and distribution standard derivation
feature_screen_sd(feature_list, sig_num = 20)feature_screen_sd(feature_list, sig_num = 20)
feature_list |
a data list, which is the output of feature_selection function |
sig_num |
the number of significant features for each cluster |
selected_features, a list the same long as the cluster number, each element is a sub-list with two vectors, one for the over-expressed features, one for the under-expressed features for the current cluster
library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) M3JF_res <- M3JF(temp_data,k=4) feature_list <- feature_selection(temp_data[[1]],M3JF_res$cluster_res,z_score=TRUE, upper_bound=1, lower_bound=-1) selected_features <- feature_screen_sd(feature_list,sig_num=20)library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) M3JF_res <- M3JF(temp_data,k=4) feature_list <- feature_selection(temp_data[[1]],M3JF_res$cluster_res,z_score=TRUE, upper_bound=1, lower_bound=-1) selected_features <- feature_screen_sd(feature_list,sig_num=20)
Select the cluster related features via hypergeometric test
feature_selection( X, clusters, z_score = FALSE, upper_bound, lower_bound, p.adjust.method = "BH" )feature_selection( X, clusters, z_score = FALSE, upper_bound, lower_bound, p.adjust.method = "BH" )
X |
the feature matrix to be analyzed, with rows as samples and columns as features |
clusters |
the numeric cluster results with number specifying the cluster |
z_score |
a binary value to specify whether to calculate z-score for X first |
upper_bound |
values larger than this value should be treated as over-expressed |
lower_bound |
values smaller than this value should be treated as under-expressed |
p.adjust.method |
the p value adjust method, defalut as 'BH' |
results, a list, which is as long as (cluster number+2), with the first (cluster number) element as two sub-list, each composing a feature vector and a FDR vector. The last two elements are two matrices, one is the matrix representing the fraction of over-express samples in each cluster for each features , and the other represents that of under-express.
library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) M3JF_res <- M3JF(temp_data,k=4) feature_list <- feature_selection(temp_data[[1]],M3JF_res$cluster_res,z_score=TRUE, upper_bound=1, lower_bound=-1)library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) M3JF_res <- M3JF(temp_data,k=4) feature_list <- feature_selection(temp_data[[1]],M3JF_res$cluster_res,z_score=TRUE, upper_bound=1, lower_bound=-1)
Initialize the shared sub-matrix E and modality specific sub-matrices list Hi
initialize_WL(WL, k)initialize_WL(WL, k)
WL |
a list of multiple modality data matrices |
k |
the cluster number |
res, a list of length N+3, where N is the number of data modality. the first N elements are the modality specific sub-matrices list Hi, the (N+1) element is the shared sub-matrix E, the last two elements are the loss defined on the shared sub-matrix E and modality specific sub-matrices list Hi.
library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4)library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4)
Generate the simulated dataset with three modalities as the work iNMF
iNMF_data_gen(Xs_dim_list=list(c(100,100),c(100,100),c(100,100)), mod_dim_list=list(matrix(c(20,30,20,30,20,30,20,30),4,2), matrix(c(20,20,30,30,20,30,20,30),4,2), matrix(c(26,24,26,24,20,30,20,30),4,2)),e_u=0.15, e_s=0.9, e_h=0)iNMF_data_gen(Xs_dim_list=list(c(100,100),c(100,100),c(100,100)), mod_dim_list=list(matrix(c(20,30,20,30,20,30,20,30),4,2), matrix(c(20,20,30,30,20,30,20,30),4,2), matrix(c(26,24,26,24,20,30,20,30),4,2)),e_u=0.15, e_s=0.9, e_h=0)
Xs_dim_list |
a list of data matrix dimensions for multiple modality data |
mod_dim_list |
a list of the dimensions of each cluster and their features |
e_u |
the level of uniform noise |
e_s |
signal to noise ratio |
e_h |
block adding probability |
res, a list of length 2, where the first element is a list of simulated data, while the second element is a vector indicating the true label of each sample.
iNMF_data <- iNMF_data_gen(Xs_dim_list=list(c(100,100),c(100,100),c(100,100)), mod_dim_list=list(matrix(c(20,30,20,30,20,30,20,30),4,2), matrix(c(20,20,30,30,20,30,20,30),4,2), matrix(c(26,24,26,24,20,30,20,30),4,2)),e_u=0.15, e_s=0.9, e_h=0)iNMF_data <- iNMF_data_gen(Xs_dim_list=list(c(100,100),c(100,100),c(100,100)), mod_dim_list=list(matrix(c(20,30,20,30,20,30,20,30),4,2), matrix(c(20,20,30,30,20,30,20,30),4,2), matrix(c(26,24,26,24,20,30,20,30),4,2)),e_u=0.15, e_s=0.9, e_h=0)
Generate the simulated dataset with three modalities with the package InterSIM
intersim_data_gen(prop=c(0.20,0.30,0.27,0.23), n_sample=500)intersim_data_gen(prop=c(0.20,0.30,0.27,0.23), n_sample=500)
prop |
proportion of samples for each cluster |
n_sample |
the number of samples |
res, a list of length 2, where the first element is a list of simulated data, while the second element is a vector indicating the true label of each sample.
library(InterSIM) intersim_data <- intersim_data_gen(prop=c(0.20,0.30,0.27,0.23), n_sample=500)library(InterSIM) intersim_data <- intersim_data_gen(prop=c(0.20,0.30,0.27,0.23), n_sample=500)
A new version of kmeans that generates stable cluster result
kmeanspp(X, k)kmeanspp(X, k)
X |
a data matrix with each row as a sample and each column as a feature |
k |
the cluster number |
res, the cluster result generated by this function
library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4) lambda <- 0.01 update_E_list <- update_E(temp_data,init_list,lambda) cluster_res <- kmeanspp(update_E_list[[4]],4)library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4) lambda <- 0.01 update_E_list <- update_E(temp_data,init_list,lambda) cluster_res <- kmeanspp(update_E_list[[4]],4)
Multi-Modal Matrix Joint Factorization
M3JF(WL, lambda = 0.01, theta = 10^-6, k)M3JF(WL, lambda = 0.01, theta = 10^-6, k)
WL |
a list of multiple modality data matrices |
lambda |
the parameter to set the relative weight of the group sparse constraint |
theta |
threshold for the stopping criteria |
k |
cluster number |
result, a list of 3 elements, the first element is a list comprising the shared sub-matrix and the modality specific sub-matrices. The second element is a vector of the clustering result. The third element is a vector of the cost in each step during optimization.
library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) M3JF_res <- M3JF(temp_data,k=4)library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) M3JF_res <- M3JF(temp_data,k=4)
Evaluate the cluster number of multiple modality data
RotationCostBestGivenGraph(W, NUMC = 2:5)RotationCostBestGivenGraph(W, NUMC = 2:5)
W |
a list of multiple modality data matrices |
NUMC |
a vector specify the data range to select best cluster number |
quality, a vector of rotation cost the same long as NUMC, where each element is the rotation cost value of the corresponding cluster number.
library(InterSIM) library(SNFtool) sim.data <- InterSIM(n.sample=100, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) dat <- lapply(temp_data, function(dd) { dd <- as.matrix(dd) dd1 <- dist2(dd,dd) W1 <- affinityMatrix(dd1, K = 10, sigma = 0.5) }) W <- SNF(dat, 10, 10) clu_eval <- RotationCostBestGivenGraph(W,2:10)library(InterSIM) library(SNFtool) sim.data <- InterSIM(n.sample=100, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) dat <- lapply(temp_data, function(dd) { dd <- as.matrix(dd) dd1 <- dist2(dd,dd) W1 <- affinityMatrix(dd1, K = 10, sigma = 0.5) }) W <- SNF(dat, 10, 10) clu_eval <- RotationCostBestGivenGraph(W,2:10)
Generate the simulated dataset with specified parameters
simulateY(nclust = 4, n_byClust = c(10,20,5,25), J=1000, prop = 0.01, noise = 0.1,flavor =c("normal", "beta", "binary"), params = list(c(mean = 1,sd = 1)))simulateY(nclust = 4, n_byClust = c(10,20,5,25), J=1000, prop = 0.01, noise = 0.1,flavor =c("normal", "beta", "binary"), params = list(c(mean = 1,sd = 1)))
nclust |
number of clusters |
n_byClust |
number of samples per cluster |
J |
number of features in each modality |
prop |
proportion of cluster related features |
noise |
percentage of noise adding to each modality |
flavor |
a vector indicating the data type |
params |
a list indicating the mean and standard derivation of the simulated data |
res, a list of length 2, where the first element is a list of simulated data, while the second element is a vector indicating the true label of each sample
temp_data <- simulateY(nclust = 4, n_byClust = c(10,20,5,25), J=1000, prop = 0.01, noise = 0.1,flavor =c("normal", "beta", "binary"), params = list(c(mean = 1,sd = 1)))temp_data <- simulateY(nclust = 4, n_byClust = c(10,20,5,25), J=1000, prop = 0.01, noise = 0.1,flavor =c("normal", "beta", "binary"), params = list(c(mean = 1,sd = 1)))
Update sub-matrix E
update_E(WL, init_list, lambda)update_E(WL, init_list, lambda)
WL |
a list of multiple modality data matrices |
init_list |
a list of the initialized modality specific sub-matrices list Hi and shared sub-matrix E |
lambda |
a parameter to set the relative weight of the L1,infinity norm defined on sub-matrices list E |
update_E_list, the data list init_list with the shared sub-matrix E updated.
library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4) update_H_list <- update_H(temp_data,init_list) lambda <- 0.01 update_E_list <- update_E(temp_data,update_H_list,lambda)library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4) update_H_list <- update_H(temp_data,init_list) lambda <- 0.01 update_E_list <- update_E(temp_data,update_H_list,lambda)
Update sub-matrices list Hi
update_H(WL, init_list)update_H(WL, init_list)
WL |
a list of multiple modality data matrices |
init_list |
a list of the initialized modality specific sub-matrices list Hi and shared sub-matrix E |
update_H_list, the data list init_list with the modality specific sub-matrices list Hi updated.
library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4) update_H_list <- update_H(temp_data,init_list)library(InterSIM) sim.data <- InterSIM(n.sample=500, cluster.sample.prop = c(0.20,0.30,0.27,0.23), delta.methyl=5, delta.expr=5, delta.protein=5,p.DMP=0.2, p.DEG=NULL, p.DEP=NULL,sigma.methyl=NULL, sigma.expr=NULL, sigma.protein=NULL,cor.methyl.expr=NULL, cor.expr.protein=NULL,do.plot=FALSE, sample.cluster=TRUE, feature.cluster=TRUE) sim.methyl <- sim.data$dat.methyl sim.expr <- sim.data$dat.expr sim.protein <- sim.data$dat.protein temp_data <- list(sim.methyl, sim.expr, sim.protein) init_list <- initialize_WL(temp_data,k=4) update_H_list <- update_H(temp_data,init_list)