dig-pigean

PIGEAN method

Usage

Uses python 3.8

pip install -r requirements.txt

base cmd:

python3 ./priors.py gibbs 
--first-for-sigma-cond 
--sigma-power -2 
--gwas-detect-high-power 100 
--gwas-detect-low-power 10 
--num-chains 10 
--num-chains-betas 4 
--max-num-iter 500 
--background-prior 0.05 
--filter-gene-set-p 0.005 
--max-num-gene-sets 4000
--min-gene-set-size 5 
--gene-loc-file ./NCBI37.3.plink.gene.loc 
--gene-map-in ./gencode.gene.map 
--gene-loc-file-huge ./refGene_hg19_TSS.subset.loc 
--exons-loc-file-huge ./NCBI37.3.plink.gene.exons.loc 
--gene-stats-out gs.out 
--gene-set-stats-out gss.out 
--gene-gene-set-stats-out ggss.out 
--gene-effectors-out ge.out

Four out files:

• gs.out (gene data),
• gss.out (gene set data),
• ggss.out (combined data),
• ge.out (gene effectors data)

Note the four required files:

• gene-loc-file: NCBI37.3.plink.gene.loc
• gene-map-in: gencode.gene.map
• gene-loc-file-huge: refGene_hg19_TSS.subset.loc
• exons-loc-file-huge: NCBI37.3.plink.gene.exons.loc

Three types of inputs can be run: GWAS sumstats, gene lists, or exome data. The inputs for each of these three are as follows:

Sumstats

--gwas-in pigean.sumstats.gz 
--gwas-chrom-col CHROM 
--gwas-pos-col POS 
--gwas-p-col P 
--gwas-n-col N

Gene Lists

--positive-controls-in gene_list.tsv 
--positive-controls-id-col 1 
--positive-controls-prob-col 2 
--positive-controls-no-header True 
--positive-controls-all-in ./refGene_hg19_TSS.subset.loc 
--positive-controls-all-no-header True 
--positive-controls-all-id-col 1

Exomes

--exomes-in exomes.sumstats.gz 
--exomes-gene-col Gene 
--exomes-p-col P-value 
--exomes-beta-col Effect

Additionally you will have to pass in a set of gene sets using the command --X-in e.g.

--X-in gene_set_list_mouse_2024.txt
--X-ing ene_set_list_string_notext_medium_processed.txt

The gene sets look like:

PPIA_string_fusion	RGPD3:0.93	RGPD8:0.92	RGPD2:0.92	RGPD1:0.92	RGPD5:0.92	RANBP1:1	RGPD4:0.93	FKBP4:0.69
PPIH_string_fusion	RGPD3:0.87	RGPD8:0.86	RGPD2:0.86	RGPD1:0.85	RGPD5:0.86	RANBP1:1	FKBP4:0.8

The probabilities (the 0.93 in RGPD3:0.93) are optional if all genes are equally weighted.