Bionexr: an R package for integrative network-based analysis of gene somatic mutation and gene expression data to identify cancer drivers

Cancer genome projects have generated massive genome and transcriptome sequencing data, which makes tumor-specific alterations such as somatic mutation and gene expression information easily available. To distinguish cancer drivers from passengers, we implement an R package “Bionexr” for integrative network-based analysis of gene somatic mutation and expression data. Bionexr provides these features:

• A protein-protein interaction (PPIN)-based approach
• A pathway-based appoach
• Visualization of the results

Bionexr is consisted of four main modules:

• Data Download
• Gene Analysis
• Network Analysis
• Visualization

<img src = "https://raw.githubusercontent.com/ys-amms/bionexr/master/fig/overview.jpg" width = "700" align = "center" />

Installation

Hardware requirements

Important: Read First

• PC with 8G RAM or above is recommended
• Passed test on Windows 8.1, Ubuntu 14.04, OS X 10.11.2

Software requirements

• Depends: R (>= 3.2.1)
• Suggests: doParallel, foreach, knitr
• Imports: RCurl, XML, reshape2, igraph, DESeq2
• VignetteBuilder: knitr

How to install

First, please install "DESeq2" package from BioConductor. To install "DESeq2", start R and enter:

source("https://bioconductor.org/biocLite.R")
biocLite("DESeq2")

Second, please install "devtools" package (see devtools github for more information). To install devtools from CRAN, please run:

install.packages("devtools")

Finnaly, follow the instructions below to download latest version of Bionexr:

devtools::install_github("ys-amms/bionexr", build_vignettes = TRUE)

Users can browse the vignette by running browseVignettes("bionexr")

Quick Start

• First, run the following instructions. Note that if it is the first time to use Bionexr, prepare_ma() will take some time to download the dataset used by gene module.

    library(bionexr)
    prepare_ma() 
  

• For PPIN-based approach, follow the instructions below:

    res.gene <- perform_gene_ppi(hnsc_mut_part, hnsc_exp_part)
    res.network <- perform_network_ppi(res.gene[[2]], res.gene[[3]])  
    g <- network_from_ppi(res.network)
    plot_ppi(g)
  

  And the result would look like below, note that your running result might have a different layout, that's OK. Mutated genes are labeled with their gene symbols, colored nodes denote differentially expressed genes, and edges represent neighbor relationship of protein-protein interaction network between genes.

<img src = "https://raw.githubusercontent.com/ys-amms/bionexr/master/fig/f1.png" width = "600" align = "center" />

• For pathway-based approach, follow the instructions below:

    res.gene <- perform_gene_pathway(hnsc_mut_part, hnsc_exp_part)
    res.network <- perform_network_pathway(res.gene[[2]], res.gene[[3]], hnsc_expressed_genes)
    g <- network_from_significant_branches(res.network)
    plot_pathway(g)
  

  And the result would look like below, note that your running result might have a different layout, that's OK. The figure shows a hierarchical network, red nodes in top layer denote mutated genes, blue nodes in middle layer denote activated transcription factors, and green nodes in bottom layer denote up-regulated target genes.

<img src = "https://raw.githubusercontent.com/ys-amms/bionexr/master/fig/f2.png" width = "600" align = "center" />

Usage

Download cancer genome data using "Data Download" module

firehose_get is the main command to download cancer genome data from firehose. Here we use firehose_get to download HNSC data.

mut_data <- firehose_get("HNSC", "mutation", run_date = "2015_08_21", run_type = "stddata")
mut_data <- mut_data[[1]]
mut_sample_ids <- unique(mut_data[[7]])
exp_data <- firehose_get("HNSC", "expression", run_date = "2015_08_21", run_type = "stddata")
exp_data <- exp_data[[1]]
exp_sample_ids <- colnames(exp_data)

common_case <- intersect(mut_sample_ids, exp_sample_ids)
exp_control <- grepl("-11$", exp_sample_ids)

hnsc_mut <- mut_data[mut_data[[7]] %in% common_case, ]
hnsc_exp <- exp_data[, (exp_sample_ids %in% common_case) | exp_control]

Using "Gene Analysis" module

perform_gene_ppi and perform_gene_pathway are the two main commands for performing "Gene Analysis". As you can guess from the function name, perform_gene_ppi is for PPIN-based approach and perform_gene_pathway is for pathway-based approach.

See the instructions below, note that hnsc_mut and hnsc_exp are from "Data Download" module:

• For PPIN-based approach

    ppi.gene <- perform_gene_ppi(hnsc_mut, hnsc_exp)
  

• For pathway-based approach

    pathway.gene <- perform_gene_pathway(hnsc_mut, hnsc_exp)
  

Note that before performing "Gene Analysis", run command prepare_ma() first. This module would take a few time to finish, drink some coffee happily.

Using "Network Analysis" module

perform_network_ppi and perform_network_pathway are the two main commands for performing "Network Analysis". As the same to "Gene Analysis" module, perform_network_ppi is for PPIN-based approach and perform_network_pathway is for pathway-based approach.

See the instructions below, note that hnsc_exp is from "Data Download" module, and ppi.gene and pathway.gene are from "Gene Analysis" module:

• For PPIN-based approach

    ppi.network <- perform_network_ppi(ppi.gene[[2]], ppi.gene[[3]])
  

• For pathway-based approach

    expressed_genes <- identify_expressed_genes(hnsc_exp)
    pathway.network <- perform_network_pathway(pathway.gene[[2]], pathway.gene[[3]], expressed_genes)
  

Using "Visualization" module

plot_ppi and plot_pathway are the two main commands for performing "Visualization" module.plot_ppi is for PPIN-based approach's result and plot_pathway is for pathway-based approach's result.

See the instructions below, note that ppi.network and pathway.network are from "Network Analysis" module:

• For PPIN-based approach's result

    ppi.g <- network_from_ppi(ppi.network)
    plot_ppi(ppi.g)
  

• For pathway-based approach's result

    pathway.g <- network_from_significant_branches(pathway.network)
    plot_pathway(pathway.g)
  

Using "Gene Analysis" and "Network Analysis" in one command

The commands perform_main_ppi and perform_main_pathway can perform "Gene Analysis" and "Network Analysis", and the result can be visualized straightforward.

The example instructions are written below, note that hnsc_mut and hnsc_exp are from "Data Download" module:

• For PPIN-based approach

    prepare_ma()  
    ppi.res <- perform_main_ppi(hnsc_mut, hnsc_exp, jobname = "HNSC", use_cache = TRUE)  
    ppi.g <- network_from_ppi(ppi.res)  
    plot_ppi(ppi.g)
  

• For pathway-based approach

    prepare_ma()  
    pathway.res <- perform_main_pathway(hnsc_mut, hnsc_exp, jobname = "HNSC", test = TRUE)  
    pathway.g <- network_from_significant_branches(pathway.res)  
    plot_pathway(pathway.g)
  

Contact

Please send email to yannis.pku@gmail.com if you have any questions.