PePr v1.1.10 or newer

Introduction

PePr is a ChIP-Seq Peak-calling and Prioritization pipeline that uses a sliding window approach and models read counts across replicates and between groups with a negative binomial distribution. PePr empirically estimates the optimal shift/fragment size and sliding window width, and estimates dispersion from the local genomic area. Regions with less variability across replicates are ranked more favorably than regions with greater variability. Optional post-processing steps are also made available to filter out peaks not exhibiting the expected shift size and/or to narrow the width of peaks.

Installation

1. Make sure your python version is higher than 2.6. Version 3.X may not be fully supported yet.
2. Install pip in your system if you don't have it.
3. pip install PePr or pip install PePr --user(if you don't have administrator privilege). Optionally, you can download tarball (PePr-[version].tar.gz) from github and install using pip install PePr-[version].tar.gz
4. If installation is successful, you could directly invoke the script by typing PePr. A help message will show up.

Supported File Formats

• Single-end: BED, BAM, SAM. For an example of BED format, please see https://genome.ucsc.edu/FAQ/FAQformat.html#format1. BED files need to have at least 6 columns (including chromsome name[1], start[2], end[3], and strand[6]. To estimate shift size (default), you need to have reads from both "+" and "-" strand".)
• Paired-end: BAM, SAM. The files must be sorted by the read names. Users can use samtools sort -n sample.bam sample.sorted_by_name to sort the file.

Scripts to call PePr

The following scripts are available after the installation and can be called directly from bash console/terminal.

• PePr: the main PePr function. Will run the entire pipline including the parameter estimation and statistical testing. This does not include post-processing step.
• PePr-preprocess: PePr preprocess function. Will only run the parameter estimation step and output a parameter file for the user to use. Users could use/modify the parameter file and then run PePr main function with it.
• PePr-postprocess: PePr postprocess function. This function will take PePr peaks and remove suspected false positives using the peak shapes. The ChIP and input files of the group where the peaks from are required for this function. For example, chip1 and input1 files will be needed for chip1_peaks. This function loads all of the reads into memory, so if you don't have enough memory on your machine, you don't have to give PePr all ChIP/input files. 2 to 3 files might suffice under certain circumstances.

Basic Usage Examples

Warning: These are working examples with minimal required parameters. For the best performance (or to avoid bad fitting) on your data, please read this manual carefully and choose the right parameters.

• For peak-calling, run:

PePr -c chip_rep1.bam,chip_rep2.bam -i input_rep1.bam,input_rep2.bam -f bam

• For differential binding analysis with input samples, run:

PePr -c chip1_rep1.bam,chip1_rep2.bam -i input1_rep1.bam,input1_rep2.bam --chip2 chip2_rep1.bam,chip2_rep2.bam --input2 input2_rep2.bam,input2_rep2.bam -f bam --diff

• For differential binding analysis without input samples, run:

PePr -c chip1_rep1.bam,chip1_rep2.bam --chip2 chip2_rep1.bam,chip2_rep2.bam -f bam --diff

• To use a parameter file, run: PePr -p parameter_file.txt. For how to write a parameter file, see the section Parameter File below.

Parameters

Parameter File

The parameter file is an easier way of running PePr by including the running parameters in one file. It is effectively the same as running from the command line. A basic example is provided below:

#filetype       filename
chip1   chip_rep1.bed
chip1   chip_rep2.bed
input1  input_rep1.bed
input1  input_rep2.bed
file-format     bed
peaktype     broad
difftest     FALSE
keep-max-dup 2
threshold     1e-5
name    test

PePr will also output a complete parameter file for you to keep a record of your running parameters and produce the same results.

Output Files

• NAME__PePr_peaks.bed: A tab-delimited file containing chromosomal position of the peak, name, signal value, fold change, p-value and Benjamini-Hochberg FDR. Peak format is same as the ENCODE BroadPeak format.
• NAME__PePr_[chip1/2]_peaks.bed: this is the same as above, but only available when you run in differential binding mode. "chip1_peaks" are enriched in chip1, "chip2_peaks* are enriched in chip2.
• NAME__PePr_parameters.txt: A file containing the parameters to reproduce the results.
• NAME-Date-Time-SessionID-debug.log: This file contains the detailed information about the running status. Useful debugging information contains: Chromosomes analyzed, shift size and window size estimation, number of candidate windows, etc.

Links

• https://github.com/shawnzhangyx/PePr/ # Source code
• https://renlab.sdsc.edu/omics/tags/PePr/ # PePr FAQ
• https://pypi.python.org/pypi/pepr #PyPI package index

Questions?

You're also welcome to shoot me an e-mail at yanxiazh@umich.edu, I'll try replying to you as soon as possible. In the e-mail, please include [1] a copy of your command/script to call PePr, [2] paramters.txt file, and [3] log file. It will speed up the troubleshooting process.

Cite PePr

Zhang Y, Lin YH, Johnson TD, Rozek LS, Sartor MA. PePr: A peak-calling prioritization pipeline to identify consistent or differential peaks from replicated ChIP-Seq data. Bioinformatics. 2014.