Skip to content

BoevaLab/DeepCAST-GWAS

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 

Repository files navigation

DeepCAST-GWAS

Running DeepCAST-FWER

Preparing the data

You will need to load your data and then adjust the configuration file src/config.py.

You will also need to provide a tracklist for the phenotype you are looking at. The tracklist name has to match the filename and should be located in the TRACKLISTS_DIR as specified in the configuration file. Support for phenotypes that are not in the PanUKBB sumstats is currently limited.

config parameters explained:

SAD_DEVIATION_FACTOR = 1

Cutoff for filtering SNPs based on their SAD scores for FWER control method.

PVAL_THRESHOLD = 5e-8

Conventional p-value threshold.

FDR_THRESHOLD = 0.0025

FDR threshold for FDR control method.

R2_THRESHOLD = 0.2 KB_RADIUS = 500 WINDOW_SIZE = 500000

Parameters for PLINK clumping.

CHROMOSOMES = list(range(1, 23))

Chromosomes to consider.

STD_BINS = [-np.inf, -2, -1, -0.5, -0.25, 0, 0.25, 0.5, 1, 2, np.inf] STD_BIN_LABELS = ["<-2σ", "[-2σ, -1σ]", "[-1σ, -0.5σ]", "[-0.5σ, -0.25σ]", "[-0.25σ, 0σ]", "[0σ, 0.25σ]", "[0.25σ, 0.5σ]", "[0.5σ, 1σ]", "[1σ, 2σ]", ">2σ"]

Bins for stratified FDR control method.

You can download the reference genome merged with the enformer tracks here.

You can download the exon regions as pdf here.

Calling the method

DeepCAST can be run by calling src/main.py (FWER control) or src/deepcast_sfdr.py (FDR control).

It has the following parameters:

--phen

Phenotype id for recomputing a PanUKBB phenotype with the provided phenotype indexing (data/phenotype_identifier.csv)

--run_id

provide a run_id to to create a result subfolder.

--filename

When running phenotypes that are not indexed, the method will look for this filename in the SUMSTATS_DIR specified in the configuration file.

--chr --bp --ref --alt --neglogp

Column names for the sumstats file can be specified in case they deviate from the deafault names specified in the configuration file.

Written with StackEdit.

About

This is the repository hosting all the source code of the DeepCAST-GWAS method.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •