Table of Contents

Introduction

PeakRanger is a multi-purporse software suite for analyzing next-generation sequencing (NGS) data. The suite contains the following tools:

1. nr noise rate estimator. Estimates signal to noise ratio which is an indicator for ChIP enrichment
2. lc library complexity calculator. Calculates the ratio of unique reads over total reads. Only accepts bam files.
3. wig coverage file generator. Generates variable step format wiggle file
4. wigpe coverage file generator. Generates bedGraph format wiggle file and supports spliced alignments and thus only supports bam files
5. ranger ChIP-Seq peak caller. It is able to identify enriched genomic regions while at the same time discover summits within these regions.
6. ccat ChIP-Seq peak caller. Tuned for the discovery of broad peaks

Both ranger and ccat supports generating HTML-based annotation reports.

wigpe can also generate coverage files for bam files containing spliced reads, such as those from RNA-Seq experiments.

If you use PeakRanger in your research, please cite:

Feng X, Grossman R, Stein L: PeakRanger:A cloud-enabled peak caller for ChIP-seq data.BMC Bioinformatics 2011, 12(1):139.

if you use the ccat tool, please also cite:

Xu, H., L. Handoko, et al. (2010).A signal-noise model for significance analysis of ChIP-seq with negative control.Bioinformatics 26(9): 1199-1204.

System Requirement

1. The ranger and ccat tool depends on the R programming environment to generate HTML reports. But they can run without it, even with --report and --gene_annot_file set.
2. When the number of peaks called by ranger or ccat is huge, it takes a while for user's browser to parse the generated HTML file.
3. The lc tool needs about 1.7G ram per 10 million aligned reads.

Obtaining PeakRanger

PeakRanger can be downloded from sourceforge.

How PeakRanger works

ranger

Calling narrow peaks

It turns out that ranger servers better as a narrow-peak caller. It behaves in a conservative but sensitive way compared to similar algorithms.

The algorithm

ranger uses a staged algorithm to discover enriched regions and the summits within them. In the first step, PeakRanger implements a FDR based adapative thresholding algorithm, which was originally proposed by PeakSeq. PeakRanger uses this thresholder to find regions with enriched reads that exceed expects. After that, PeakRanger searches for summits in these regions. The summit-search algorithm first looks for the location with largest number of reads. It then searchs for sub-summits with the sensitivity, the delta -r, specified by the user. Smaller -r will generate more summits.The coverage profiles are smoothed and padded before calling summits. The smoothing grade varies with -b. Higher smoothing bandwidth results less false summits at the cost of degraded summit accuracy .To measure the significance of the enriched regions, PeakRanger uses binormial distribution to model the relative enrichment of sample over control. A p value is generated as a result. Users can thus select highly significant peaks by using a smaller -p.

Reads extending

ranger extends reads before calling peaks. The default reads extension length is 200. However, users can change this by -l if the datasets come with a different fragment size. The extension length will change the reads coverages generated from the raw reads as it will change the heights of peaks.

wigpe and wig

To help visualizing the results, wigpe and wig generates reads coverage files in the wig format. These files can then be loaded into browsers to evaluate the authenticity of called peaks. Since smaller wiggle files take less time and memory to load, --split can be set to generate one small wig file per chromosome.

ccat

Calling broad peaks

Calling broad peaks remain unsolved for the ChIP-Seq community. It seems the CCAT algorithm is one of those that is designed for this problem, especially for calling histone modification marks.

The algorithm

For details of the algorithm, please refer to the original manuscript of CCAT:

Xu, H., L. Handoko, et al. (2010).A signal-noise model for significance analysis of ChIP-seq with negative control.Bioinformatics 26(9): 1199-1204.

nr

nr is a module of the original CCAT algorithm that estimates the similarity of data and control. It indicates roughly how data departs from control

lc

lc measures the percentage of unique reads. The result measures how diversified the reads are in the dataset. The idea is from:

Chen, Yiwen, Nicolas Negre, Qunhua Li, Joanna O. Mieczkowska, Matthew Slattery, Tao Liu, Yong Zhang, et al. 2012. Systematic evaluation of factors influencing ChIP-seq fidelity. Nature Methods 9(6): 609-614.

Compiling PeakRanger from source codes

Compiling in Ubuntu

Required libraries before compiling:

1. The Boost library v1.47 or newer

2. Pthread

3. g++

Once all the libraries are installed, go to the root path of the unzipped package and type:

make

This will generate bin/peakranger. Compilation in other Linux distributions is similar.

Compiling in Mac OSX

Required libraries before compiling:

1. Xcode developer tool kit from Apple

2. The Boost library v1.47 or newer

The Xcode kit can be installed using the OSX installation disk. If you dont have the installation disk, you can also get it for free from Apple Developer. The tool kit installs essential command line tools such as make and C++ compilers. The Boost library can be installed by following the instructions on its website. If you do not have root access, add the BOOST_PATH variable to the make file:

BOOST_PATH = -I/path/to/your/boost/header -L/path/to/your/boost/library

and add it to the variables of g++.

Once all the libraries are installed, go to the root path of the unzipped package and type:

make

If the compilation failed, double check the BOOST_PATH variable is correctly set. The resulting binaries require dynamic boost library files, to make sure peakranger can find these files, type :

export DYLD_LIBRARY_PATH=$DYLD_LIBRARY_PATH:/path/to/your/boost/library

please change the path accordingly.

Compiling in Windows

Not supported but should be possible.

Synopsis

peakranger lc sample.bam

peakranger nr --format bam sample.bam control.bam

peakranger wig --format bam sample.bam sample.bam_coverage

peakranger wig --format bed sample.bed sample.bed_coverage

peakranger wigpe sample.bam sample.bam_coverage

peakranger wigpe sample.bam sample.bam_coverage_splitted -s

peakranger wigpe sample.bam sample.bam_coverage_splitted_by_strand -sx

peakranger wigpe sample.bam sample.bam_coverage_gzipped -z

peakranger wigpe sample.bam sample.bam_coverage_splitted_by_strand_gzip -sxz

peakranger wigpe sample.bam sample.bam_coverage_read_extend_to_200 -l 200

peakranger ranger --format bam sample.bam control.bam ranger_result

peakranger ranger --format bam sample.bam control.bam ranger_result_threaded_faster -t 3

peakranger ccat --format bam sample.bam contro.bam ccat_result

peakranger ccat --format bam sample.bam contro.bam ccat_result_with_HTML_report
--report --gene_annot_file hg19refGene.txt

peakranger ccat --format bam sample.bam contro.bam ccat_result_with_HTML_report_5kb_region
--report --gene_annot_file hg19refGene.txt
--plot_region 10000

Command line options

nr

input

Qualities

Other

lc

input

Other

wig

input

Output

Qualities

Other

wigpe

input

Output

Qualities

Other

ranger

input

Output

Qualities

Running modes

Other

ccat

input

Output

Qualities

Other

Output files

lc

lc does not generate any files

nr

nr does not generate any files

wig

wig generates a single file by default. When -x or -s is specified, it generates multiple files depending on the datasets.

wigpe

similar to wig

ranger

Three files will be geneated:

_summit.bed

_region.bed

_details

The first two bed files can be visualized in IGV. _summit.bed file contains the locations of summits ranked by their FDR. _regions.bed file contains the locations of regions ranked by their FDR. Each summit or region is annotated by the 4th column.

_details file contains both summits and regions as well as the regions's FDR and p values.

When --report is enabled, this file will also contain nearby genes of called peaks.

--report enables HTML reporting that generates a folder named using the data file's name. The folder contains a single index.html visualizable in most browsers.

ccat

Similar to ranger