ld-clump

This is a nextflow pipeline which takes as input data from eQTLGen, BGEN files for a cohort-of-interest, and runs LD clumping for a set of variants associated with a specific Gene Symbol. The purpose of this is to generate a final list of eQTLs that are independent. This is necessary due to the fact that the input from eQTLGen are cis-eQTLs which, by design, are close together and likely to be in LD with one another.

Setup

Software requirements

To run this pipeline, you must have nextflow and singularity installed. To install these individually, please see https://www.nextflow.io/docs/latest/getstarted.html#installation and https://docs.sylabs.io/guides/3.0/user-guide/installation.html#installation. This can also be done through conda using the env.yaml file contained in this repository. Please install conda on your operating system following the instructions here: https://docs.anaconda.com/miniconda/#quick-command-line-install.

Once conda is installed, you can create the environment using the following command:

conda env create --file env.yaml

Once installed, please activate your environment so that it can be used to run the pipeline.

conda activate nextflow

Configuration

The configuration for a given run is specified in a run-specific config file, which you create and is dataset-specific. An example is shown in genomicc.config. The following parameters are required:

• BGEN_FILES [ required ] : absolute path to the BGEN files for your cohort-of-interest. This should be specified using brace expansion for each chromosome as well as each file type (bgen, bgen.bgi, sample).
• ASSEMBLY [ required ] : either grch37 or grch38 (also will accept hg19 or hg38), which is used to define problem areas of the genome to exclude (See: https://github.com/gabraham/flashpca/blob/master/exclusion_regions_hg19.txt).
• BAALNF_OUTDIR [ required ] : Output directory that contains the output from running the baal-nf pipeline. There should be one file in here per-TF.
• eQTLGEN_DATA [optional, default: 2019-12-11-cis-eQTLsFDR-ProbeLevel-CohortInfoRemoved-BonferroniAdded.txt from eQTLGen]: Downloaded data from eQTLGen, in the form of a tab-separated text file. This contains all SNPs you would like to run LD clumping on, which are associated with a given transcription factor. This should be formatted the same as the output from eQTLGen (See: https://molgenis26.gcc.rug.nl/downloads/eqtlgen/cis-eqtl/2019-12-11-cis-eQTLsFDR-ProbeLevel-CohortInfoRemoved-BonferroniAdded.txt.gz). The SNP IDs must be in the same format as the IDs in your BGEN_FILES. If this is not provided, the file will be automatically pulled from eQTLGen.
• R2_THRESHOLD [optional, default: 0.8] : R-squared threshold to use for idenftifying which SNPs are in LD with one another. Default is 0.8.
• INFO_THRESHOLD [ optional, default: 0.9 ] : Info score threshold set on imputation quality. Anything below this value will not be considered for hard-calling genotypes from the BGEN files. The default is 0.9.
• MAF_THRESHOLD [ optional, default: 0.01 ] : Minor allele frequency below which SNPs will be filtered out from your final estimands file.
• OUTDIR [ optional, default: output ] : Output directory for results.
• SNPSTATS_CACHE [ optional, default: OUTDIR/info_scores ] : Cache for saving snp stats output from QCtoolv2. This contains the output from the process bgen::generate_info_score(), and includes the following crucial information for every SNP in your BGEN_FILES: rsid, info, minor_allele_frequency.

Run

Run profiles

The following profiles are available in this pipeline:

• eddie
• ultra2

To run this pipeline, just run the following command:

nextflow run main.nf -profile your_profile -c path/to/custom.config -with-trace

Docker containers

Docker containers for this workflow can be found either in the containers directory in this repository, or at https://github.com/TARGENE/ld-block-removal/containers