ARCHIE (Aggregative trans-association analysis to identify trait-specific target genes)

Manuscript: Novel Aggregative trans-eQTL Association Analysis of Known Genetic Variants Detect Trait-specific Target Gene-sets. Diptavo Dutta, Yuan He, Ashis Saha, Marios Arvanitis, Alexis Battle, Nilanjan Chatterjee.

Preprint can be found on medrxiv

Installation and pre-requisites

The codes in this repository are writted in R. To run them, please download and install the following R packages: optparse and data.table.

Running instructions

We have provided an example run of ARCHIE for demonstration purposes. The individual application and customization would critically depend on the particular application. We encourage the user to closely follow the outlined algorithm in the preprint to adapt it to their cases.

The repo can be cloned as:

git clone https://github.com/diptavo/ARCHIE/

We have provided an example dataset to demonstrate the functionality of ARCHIE. The simulated dataset contains 100 independent (no LD) SNPs that are significantly (GWAS) associated with a phenotype and 1000 independent genes. The simulation setting was such that SNPs 1 to 5 are associated with genes 1 to 10; SNPs 11 to 15 are associated with genes 11 to 20. Three example input files for Sigma_GE, Sigma_GG and Sigma_EE are provided.

Run ARCHIE using the following example command:

Rscript ~/archie/codes/run_archie.R --sigmage ~/archie/files/sigmage.txt --sigmagg ~/archie/files/sigmagg.txt --sigmaee ~/archie/files/sigmaee.txt

This will produce out.rds.

> a1 = readRDS("out.rds");
> a1$q
0.07748791 0.02815841

> head(a1$us,20)
      [,1] [,2]
 [1,]    0    1
 [2,]    0    1
 [3,]    0    1
 [4,]    0    1
 [5,]    0    1
 [6,]    0    0
 [7,]    0    0
 [8,]    0    0
 [9,]    0    0
[10,]    0    0
[11,]    1    0
[12,]    1    0
[13,]    1    0
[14,]    1    0
[15,]    1    0
[16,]    0    0
[17,]    0    0
[18,]    0    0
[19,]    0    0
[20,]    0    0

head(a1$vs,20)
      [,1] [,2]
 [1,]    0    1
 [2,]    0    1
 [3,]    0    1
 [4,]    0    1
 [5,]    0    1
 [6,]    0    1
 [7,]    0    1
 [8,]    0    1
 [9,]    0    1
[10,]    0    1
[11,]    1    0
[12,]    1    0
[13,]    1    0
[14,]    1    0
[15,]    1    0
[16,]    1    0
[17,]    1    0
[18,]    1    0
[19,]    1    0
[20,]    1    0

us and vs in the output represent the selected SNPs mapped to corresponding selected genes (1 indiciating selected, 0 indicating not selected). In the output, note that SNPs 1 to 5 has been correctly mapped to genes 1 to 10 and SNPs 11 to 15 mapped to genes 11 to 20. ARCHIE usually numerically determines the number of sparse canonical correlation components to be extracted, however that is customizable as well. q is the aggregated measure of association explained by the corresponding component.

Detailed options

To get the detailed options for ARCHIE use the following command:

Rscript ~/archie/codes/run_archie.R --help

Results

trans-association results from the summary statistics of eQTLGen analysis for 29 traits will be available shortly.

Questions

Please direct any questions/suggestions/comments to diptavo21@jhu.edu

Logs

(Nov 8/2020): Updated initial release.