README.md

In Queen0044/HiCImpute: HiCImpute: An R Package Implementing "Bayesian Hierarchical Model for Identifying Structural Zeros and Enhancing Single-cell Hi-C Data"

HiCImpute

HiCImpute (Xie, Han, Jin, and Lin, 2021) is a Bayesian hierarchy model that goes beyond data quality improvement by also identifying observed zeros that are in fact structural zeros. HiCImpute takes spatial dependencies of scHi-C 2D data structure into account while also borrowing information from similar single cells and bulk data, when such are available.

Installation

The HiCImpute package has the following R-package dependencies: Rcpp, RcppArmadillo, parallel, Rtsne, ggplot2, ggpubr, and mclust. The dependent packages will be automatically installed along with HiCImpute. You can use the following commands to install HiCImpute from GitHub.

# Install and load "devtools" package. 
install.packages("devtools")
library("devtools")

# Install "HiCImpute" package from github.
install_github("Queen0044/HiCImpute")

If you are Windows user, please install Rtools40 (https://cran.r-project.org/bin/windows/Rtools/) first, and restart R to install HiCImpute package. Note that Rtools40 is for R 4.0.0+ so that you might have to update your R version.

If you have OneDrive backing-up “C:\User\Your_user_name\Documents”, the installation may fail. You can download the zip file from Github and install HiCImpute manually.

Example

This is a basic example which shows you how to solve a common problem:

library(HiCImpute)
#' data("K562_T1_7k")
#' data("K562_bulk")
#' T1_7k_res=MCMCImpute(scHiC=K562_T1_7k,bulk=K562_bulk,
#' startval=c(100,100,10,8,10,0.1,900,0.2,0,replicate(dim(scHiC)[2],8)),n=61,mc.cores = 1,
#' cutoff=0.5, niter=100000,burnin=5000)

For more information of functions, please read the vignettes.

Queen0044/HiCImpute documentation built on Oct. 9, 2022, 9:30 a.m.