DEbPeak - Analyze and integrate multi-omics to unravel the regulation of gene expression.

Introduction

DEbPeak aims to explore, visualize, interpret multi-omics data and unravel the regulation of gene expression by combining RNA-seq with peak-related data (eg: ChIP-seq, ATAC-seq, m6a-seq et al.). It contains eleven functional modules:

• Parse GEO: Extract study information, raw count matrix and metadata from GEO database.
• Quality Control (QC): QC on count matrix and samples.
  • QC on count matrix: Proportion of genes detected in different samples under different CPM thresholds and the saturation of the number of genes detected.
  • QC on samples: Euclidean distance and pearson correlation coefficient of samples across different conditions, sample similarity on selected principal components (check batch information and conduct batch correction) and outlier detection with robust PCA.
• Principal Component Analysis (PCA): this module can be divided into three sub modules, basic info, loading related and 3D visualization.
  • Basic info: scree plot (help to select the useful PCs), biplot (sample similarity with corresponding genes with larger loadings) and PC pairs plot (sample similarity under different PC combinations).
  • Loading related: visualize genes with larger positive and negative loadings on selected PCs, conduct GO enrichment analysis on genes with larger positive and negative loadings on selected PCs.
  • 3D visualization: visualize samples on three selected PCs.
• Differential Analysis and Visualization: this module includes seven powerful visualization methods (Volcano Plot, Scatter Plot, MA Plot, Rank Plot, Gene/Peak Plot, Heatmap, Pie Plot for peak-related data).
• Functional Enrichment Analysis (FEA): GO enrichment analysis, KEGG enrichment analysis, Gene Set Enrichment Analysis (GSEA).
  • GO (Biological Process, Molecular Function, Cellular Component) and KEGG on differential expression genes or accessible/binding peaks.
  • GSEA on all genes (Notice: GSEA is not available for peak-related data)
• Predict transcription factors (PredictTFs): Identify transcription factors with differentially expressed genes, DEbPeak provides three methods (BART, ChEA3 and TFEA.ChIP).
• Motif analysis:
  • de novo motif discovery
  • motif enrichment
• Integrate RNA-seq with peak-related data:
  • Get consensus peaks: For multiple peak files, get consensus peaks; for single peak file, use it directly (used in consensus integration mode).
  • Peak profile plots: Heatmap of peak binding to TSS regions, Average Profile of ChIP peaks binding to TSS region, Profile of ChIP peaks binding to different regions (used in consensus integration mode).
  • Peak annotaion (used in consensus integration mode).
  • Integrate RNA-seq with peak-related data (consensus mode): Integrate RNA-seq with peak-related data to find direct targets, including up-regulated and down-regulated.
  • Integrate RNA-seq with peak-related data (differential mode): Integrate RNA-seq and peak-related data based on differential analysis.
  • Integration summary: include venn diagram and quadrant diagram (differential mode).
  • GO enrichment on integrated results.
  • Find motif on integrated results: Due to the nature of ATAC-seq, we usually need to find motif on integrated results to obtain potential regulatory factors.
• Integrate RNA-seq with RNA-seq:
  • Integration summary: include venn diagram and quadrant diagram.
  • GO enrichment on integrated results.
• Integrate peak-related data with peak-related data:
  • Integration summary: include venn diagram and quadrant diagram (differential mode).
  • GO enrichment on integrated results.
• Utils: useful functions, including creating enrichment plot for selected enrichment terms, gene ID conversion and count normalization(DESeq2’s median of ratios, TMM, CPM, TPM, RPKM).

To enhance the ease of use of the tool, we have also developed an web server for DEbPeak that allows users to submit files to the web page and set parameters to get the desired results. Unlike the standalone R package, the web server has built-in DESeq2 for differential analysis, while the R package can accept user input results from DESeq2 or edgeR, which will be more flexible.

By the way, all plots generated are publication-ready , and most of them are based on ggplot2, so that users can easily modify them according to their needs. We also provide various color palettes, including discrete and continuous, color blind friendly and multiple categorical variables.

Framework

Application scenarios for multi-omics integration

Installation

R package

You can install the package via the Github repository:

# install.packages("devtools")   #In case you have not installed it.

# install prerequisites for enrichplot and ChIPseeker
devtools::install_version("ggfun", version = "0.0.6", repos = "https://cran.r-project.org")
devtools::install_version("aplot", version = "0.1.6", repos = "https://cran.r-project.org")
devtools::install_version("scatterpie", version = "0.1.7", repos = "https://cran.r-project.org")

# For mac, you may need to install xquartz: brew install --cask xquartz

# install DEbPeak
devtools::install_github("showteeth/DEbPeak")

In general, it is recommended to install from Github repository (update more timely).

For other issues about installation, please refer Installation guide.

Install additional tools:

# install MSPC --- consensus peak
wget --quiet https://github.com/Genometric/MSPC/releases/latest/download/linux-x64.zip -O MSPC_linux_x64.zip && unzip -q MSPC_linux_x64.zip -d mspc && cd mspc && chmod +x mspc

# install meme --- motif anaysis
## install from source
cd /opt && wget --quiet https://meme-suite.org/meme/meme-software/5.5.5/meme-5.5.5.tar.gz -O meme-5.5.5.tar.gz && tar -zxf meme-5.5.5.tar.gz && cd meme-5.5.5 && ./configure --prefix=`pwd`/meme-5.5.5/meme --enable-build-libxml2 --enable-build-libxslt && make && make install
## install from conda: conda install -c bioconda meme

# install homer --- motif enrichment
## install from source
mkdir homer && cd homer && wget --quiet http://homer.ucsd.edu/homer/configureHomer.pl -O configureHomer.pl && chmod +x configureHomer.pl && perl configureHomer.pl -install
## install from conda: conda install -c bioconda homer
## Downloading Homer Packages: http://homer.ucsd.edu/homer/introduction/install.html

# install deeptools and bart
pip install deeptools numpy pandas scipy tables scikit-learn matplotlib
wget --quiet https://virginia.box.com/shared/static/031noe820hk888qzcxvw1cazol1gdhi0.gz -O bart_v2.0.tar.gz && tar -zxf bart_v2.0.tar.gz
## Download the resources and setup the configuration file
## https://zanglab.github.io/bart/index.htm#install

Docker

We also provide a docker image to use:

# pull the image
docker pull soyabean/debpeak:1.2

# run the image
docker run --rm -p 8888:8787 -e PASSWORD=passwd -e ROOT=TRUE -it soyabean/debpeak:1.2

Notes:

• After running the above codes, open browser and enter http://localhost:8888/, the user name is rstudio, the password is passwd (set by -e PASSWORD=passwd)
• If port 8888 is in use, change -p 8888:8787
• The meme suit path: /opt/meme-5.5.5/meme/bin.
• The homer suit path: /opt/homer/bin.
• The configureHomer.pl path: /opt/homer.
• The bart path: /opt/bart_v2.0/bin
• You still need to download the resources and setup the configuration file for bart and download species packages for homer.

Usage

Vignette

Detailed usage is available in here. We divide these vignettes into four categories:

• For parse GEO:

  • Parse GEO

• For analyzing RNA-seq:

  • Quality Control
  • Principal Component Analysis (RNA-seq)
  • Differential Expression Analysis (RNA-seq)
  • Functional Enrichment Analysis (RNA-seq)
  • Predict Transcription Factors (RNA-seq)
  • Utils

• For analyzing peak-related data:

  • Quality Control
  • Principal Component Analysis (Peak-related)
  • Differential Analysis (Peak-related)
  • Functional Enrichment Analysis (Peak-related)
  • Motif Analysis (Peak-related)

• Integrating RNA-seq (differential expression analysis) with peak-related data (consensus peak):

  • Integrate RNA-seq with ChIP-seq (consensus peak)
  • Integrate RNA-seq with ATAC-seq (consensus peak)
  • Integrate RNA-seq ChIP-seq and ATAC-seq

• Integrating RNA-seq (differential expression analysis) with peak-related data (differential accessible/binding analysis):

  • Integrate RNA-seq with ATAC-seq (differential analysis)
  • Integrate RNA-seq with ChIP-seq (differential analysis).

Function list

Notice

• The KEGG API has changed, to perform KEGG enrichment, you'd better update clusterProfiler >= 4.7.1.

Contact

For any question, feature request or bug report please write an email to songyb0519@gmail.com.

Code of Conduct

Please note that the DEbPeak project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.