greenelab / tad_pathways_pipeline

Pipeline to implement a "TAD_Pathways" analysis. Discover candidate genes based on association signals in TADs

TAD_Pathways

Leveraging TADs to identify candidate genes at GWAS signals

Gregory P. Way, Casey S. Greene, and Struan F.A. Grant - 2017

Summary

The repository contains data and instructions to implement a "TAD_Pathways" analysis for over 300 different trait/disease GWAS or custom SNP lists.

TAD_Pathways uses the principles of topologically association domains (TADs) to define where an association signal (typically a GWAS signal) can most likely impact gene function. We use TAD boundaries as defined by Dixon et al. 2012 and hg19 Gencode genes to identify which genes may be implicated. We then perform an overrepresentation pathway analysis to identify significantly associated pathways implicated by the input TAD-defined geneset.

For more specific details about our method, refer to our short report at the European Journal of Human Genetics.

We also present a 6 minute video introducing the method and discussing the experimental validation at EJHG-tube.

Setup

First, clone the repository and navigate into the top directory:

git clone [email protected]:greenelab/tad_pathways_pipeline.git
cd tad_pathways_pipeline

Before you begin, download the necessary TAD based index files and GWAS curation files and setup python environment:

bash initialize.sh

# Using conda version 4.4.11
conda activate tad_pathways

Now, a TAD_Pathways analysis can proceed. Follow an example pipeline to work from an existing GWAS or the custom pipeline example for insight on how to run TAD_Pathways on user curated SNPs.

Examples

We provide three different examples for a TAD pathways analysis pipeline. To run each of the analyses:

source activate tad_pathways

# Example using Bone Mineral Density GWAS
bash example_pipeline_bmd.sh

# Example using Type 2 Diabetes GWAS
bash example_pipeline_t2d.sh

# Example using custom input SNPs
bash example_pipeline_custom.sh

General Usage

There are two ways to implement a TAD_Pathways analysis:

1. GWAS
2. Custom

GWAS

To perform a TAD_Pathways analysis on publicly available GWAS results, simply browse the data/gwas_catalog/ directory to select a valid GWAS file. These files contain a curation of all significant SNPs mapped to specific traits as distributed by the NHGRI-EBI GWAS Catalog.

Each file in this directory is a tab separated text file of genome-wide significant SNPs and their genomic location along with their reported nearest gene and associated PUBMED id. For complete information on how these files were constructed, refer to https://github.com/greenelab/tad_pathways.

Each GWAS has 3 associated files, including files in data/gwas_catalog/. The other files are located in data/gwas_tad_snps/ and data/gwas_tad_genes/. All files are important for performing a TAD_Pathways analysis. See the GWAS example files for instructions on how to implement the necessary scripts.

Custom

To perform a TAD_Pathways analysis on a list of custom SNPs, generate a comma separated text file. The first row of the text file should have group names and subsequent rows should list the rs numbers of interest. There can be many columns with variable length rows.

E.g.: custom_example.csv

Group 1	Group 2
rs12345	rs67891
rs19876	rs54321
...	...

Then, perform the following steps:

source activate tad_pathways

# Map custom SNPs to genomic locations
Rscript --vanilla scripts/build_snp_list.R \
        --snp_file "custom_example.csv" \
        --output_file "mapped_results.tsv"

# Build TAD based genelists for each group
python scripts/build_custom_TAD_genelist.py \
       --snp_data_file "mapped_results.tsv" \
       --output_file "custom_tad_genelist.tsv"

The output of these steps are Group specific text files with all genes in TADs harboring an input SNP. See example_pipeline_custom.sh for more details.

Contact

For all questions and bug reporting please file a GitHub issue

For all other questions contact Casey Greene at [email protected] or Struan Grant at [email protected]

tad_pathways_pipeline's Issues

Fix Chromosome Patches in build_snp_list.R

https://www.biostars.org/p/106355/

Validate tad_pathways config file

in #34 @dongbohu suggested adding error handling to tad_pathways/config.py. Adding this issue so that I remember to add this handling.

checkpoint in "build_snp_list.R"

Hi gwaygenomics!
I found the checkpoint usd in "build_snp_list.R" is "2016-02-25", but in "install.R" the R libraries are installed under "2017-05-22". So I met some problems as can't load some libraries. Should it be "2017-05-22" in "build_snp_list.R"?

Thanks!

Move from WebGestalt to Custom Implementation

GSEApy seems like a better alternative to WebGestalt. The primary reason is that GSEApy is on anaconda and WebGestalt has been very difficult to get working with conda (conda-forge/staged-recipes#5215, bioconda/bioconda-recipes#4795)

Move webgestalt steps to webgestaltR

I will explore moving all manual steps to programmatic steps through WebGestaltR

Transition to Notebook Style Analyses

Now that the WebGestalt step can be automated, the pipeline will benefit from a transition to notebook style analyses.

Currently, there are two ways to interact with a TAD_Pathways analysis:

1. Input Trait/GWAS
2. Custom SNP list input

For the first option, all of the analyses can be precompiled and presented as individual notebooks with analysis results and visualizations.

For the second option, the user can interact with an example notebook to input their own custom derived SNP list.

chr missing annotations

Some genes are missing chr annotations in *_gene_evidence_summary.tsv file:

gene	evidence	TAD ID	chromosome	TAD Start	TAD End	UCSC
ZCCHC12	mut	3028	chr	117555972	118355972	chr:117555972-118355972
CD99L2	mut	3055	chr	149929342	150249342	chr:149929342-150249342
FOXO4	mut	2986	chr	70203275	70923275	chr:70203275-70923275
OGT	tad	2986	chr	70203275	70923275	chr:70203275-70923275
ZIC3	mut	3045	chr	136172334	137132334	chr:136172334-137132334
POF1B	mut	2997	chr	84353344	84833344	chr:84353344-84833344
ZXDA	mut	2975	chr	56783275	58582012	chr:56783275-58582012