Note: This repository is a frozen snapshot of the software version published in Bioinformatics (doi: 10.1093/bioinformatics/btw169)[http://bioinformatics.oxfordjournals.org/content/32/15/2256.long]. For the latest developmental version, see https://github.com/mdozmorov/genome_runner/tree/shiny
GenomeRunner web is a tool for investigation of potential functional impact of sets of single nucleotide polymorphisms (SNPs) by considering their co-localization with fgenome annotation data (regulatory datasets). The philosophy behind GenomeRunner is that SNPs are not acting in isolation and may collectively alter regulatory/epigenomic features. Finding which regulatory features are affected may help to understand mechanisms of complex diseases from a holistic perspective.
GenomeRunner calculates enrichment p-values (Fisher's exact test) by evaluating whether a set of SNPs co-localizes with regulatory datasets more often that could happen by chance. GenomeRunner further performs three interpretation-oriented analyses:
- Regulatory similarity analysis, aimed at identifying groups of SNP sets having similar functional impact;
- Differential regulatory analysis, developed to identify regulatory signatures differentially enriched between the groups of SNPs;
- Cell type regulatory enrichment analysis, designed to identify cell specificity of the regulatory enrichments.
An example of GenomeRunner’s results can be found in the analysis of Sjogren’s syndrome GWAS (Nature Genetics ), where it identified RFX5 transcription factor binding site as the most statistically significantly co-localized with the set of disease-associated SNPs.
Installation instructions
sudo ./setup.shwill install all required packages. See the main documentation on https://mdozmorov.github.io/grdocs/index.html
Use R.GenomeRunner to set up Shiny server, see tips
gr-server -g hg19 -d /path/to/database/ -r /path/to/database/ -w 1 -p 8080The "-g" argument specifies organism, the "-d" argument specifies path to the database of genome annotation data, the "-r" argument specifies path to the folder to output results, the optional "-w" argument specifies number of workers to run parallel jobs, the optional "-p" argument specifies at which port to start the server. Access the server locally at "http://localhost:8080"
Prerequisites:
- Database of genome annotations, created with
gr-dbcreator. Use full path. Example: /home/genomerunner/db_5.00_07-22-2015; - Text file with full path(s) to BED file(s) containing genomic coordinates of SNPs of interest. Example: "foi.txt", where each line contains paths to BED files;
- Text file with full paths to genome annotation BED files. Example: "gf_encTfbs.txt", where each line contains paths to TFBS genome annotation BED files;
- The "background" file, a BED file containing genomic coordinates of all SNPs. This file is used to estimate enrichments that can happen by chance. The SNPs of interest should be a subset of this file.
The files containing categories of genome annotations can be created with
DIR=/home/genomerunner/db_5.00_07-22-2015
for file in `find $DIR/grsnp_db/hg19/ENCODE -maxdepth 1 -mindepth 1 -type d`; do
gf=`basename $file`; echo $gf;
find `$file -type f -name "*.bed.gz"` | sort > gf_enc$gf.txt;
doneThe logic here is that the first find command finds top folders containing categories of genomic annotation data. The second find command outputs paths for each genome annotation data file into a category-specific file name, e.g., gf_encTfbs.txt.
The analysis can be run as, e.g.,
gr -g hg19 -d /path/to/database -r /path/to/results/folder -a fois.txt gf_encTfbs.txt /path/to/background.bed The optional "-a" argument specify if annotation analysis should be run.
This release is manually sunchronized with the developmental branch of GenomeRunner using clear; diff -r --brief genome_runner/ genomerunner_web/ | grep differ | grep -v .git | sort and clear; diff -r --brief R.GenomeRunner/ genomerunner_web/R.GenomeRunner/ | grep differ | grep -v .git | sort