Variance Decomposition using LIMIX

Snakemake pipeline to perform variance decomposition using LIMIX.

• GitHub repo: https://github.com/CollinsLabBioComp/flexible_variance_decomposition
• Free software: MIT license

Description

This pipeline is designed to perform variance decomposition in molecular trait (e.g., RNA-seq) data using LIMIX v3.0.4. Briefly, we calculate a covariance matrix from a set of observations (e.g., genetics, differential potential, etc.) and model the matrix as a random effect in LIMIX to estimate the proportion of variance in the molecular trait explained by the effect. As input, it takes molecular trait data and either (1) a data frame of values to convert to a covariance matrix or (2) a pre-calculated covariance matrix.

For detailed description of LIMIX: LIMIX: genetic analysis of multiple traits

Quickstart

Quickstart for deploying this pipeline locally and on a high performance compute cluster.

1. Set up the environment

See environment README to set up environment. Once the environment is set up, activate the conda environment:

source activate limix-vardec

Alternatively, if using singularity or docker, one can pull the image from henryjt/flexible_variance_decomposition:1.0.0.

2. Prepare the input files

Generate and/or edit input files for the pipeline.

As input, the pipeline expects the following files in the data/ directory in the location you are running the pipeline:

1. moltraits.tsv.gz: A TSV file containing molecular trait data where samples are columns and rows are features. All metadata information should be before the samples start. For example:

   chr	start	end	gene	sample_1	sample_2	...	sample_n
   chr11	2159779	2161221	INS	40.5	241.5	...	591.1
   chr2	162142882	162152404	GCG	72.5	10.5	...	1000.1

2. samples.txt: Text file containing all samples to use for analysis. Each line should be a sample ID and should be contained within moltraits.tsv.gz and covariates.tsv.gz
3. covariates.tsv.gz: TSV file containing covariates to include in the model. First column should correspond to sample IDs.

Examples of these files can be found in demo/.

3. Run pipeline

NOTE: All input file paths should be full paths.

To run:

snakemake \
   --snakefile "/path/to/repo/dir/Snakefile" \
   --configfile "/path/to/config/config_analysis.json"

Examples:

• Local with docker
• Local with conda
• Cluster with conda

Notes

• The primary LIMIX documentation is no longer being supported. To find more information, see the temporary documentation.

Authors: Henry Taylor