PLINK Association Tests - ohsu-comp-bio/cedar-gwas GitHub Wiki

Convert Minimac3 VCF to PLINK

PLINK has its own file formats including: .bim, .bed, .fam. Note the .bed file is not the same as UCSC's BED format. Example conversion:

plink --vcf [minimac3 vcf] --out [plink prefix]

Decide best association test

PLINK comes with its own suite of association tests to choose from, such as: Case/Control, Fisher's Exact Test, Cochran-Armitage test, linear/logistic models. Docs for plink association tests.

You can also write your own Rscript to perform a custom test. Example here. Docs for using the PLINK Rplugin.

Run PLINK

To run a plink association test, you need the plink binary files, Rscript with chosen association test (optional), and a phenotype file (optional). The phenotype file contains 3 columns: Family ID, Individual ID, Phenotype. Example:

TCGA-EJ-5499    TCGA-EJ-5499    0.198680431181652
TCGA-EJ-5498    TCGA-EJ-5498    0.083903480054661
TCGA-CH-5772    TCGA-CH-5772    0.207180609292677

It's also a good idea to filter out variants that have a minor allele of less than 10% (--maf 0.1) and fail hardy-weinberg test (--hwe 1e-8).

Example Command using Spearman Correlation (see wiki for using the R-plugin):

plink --bfile [prefix to plink files] --R Rplink.spearman.correlation.R --pheno [phenotype file] --maf 0.1 --hwe 1e-8 --out [prefix for output files] --noweb --allow-no-sex

Example Command using Case/Control Association:

plink --bfile [prefix to plink files] --pheno [phenotype file] --assoc --noweb --maf 0.1 --hwe 1e-8 --out [prefix for output files] --allow-no-sex

Output

The output includes information about the SNP/Allele being tested, the Spearman coefficient, p-value, and confidence intervals.

CHR SNP BP  Allele  Spearman    P   LCI UCI
10  rs376835509 65030   A   0.0042604   0.938356    -0.103425   0.111847
10  rs72635988  69083   C   -0.0555186  0.313186    -0.162187   0.0524324