PopCluster

PopCluster: a new algorithm to identify genetic variants with ethnicity-dependent effects

Dependencies

• Before running the program, make sure that R, PLINK, and EIGENSOFT (version 6.1.4) are installed.
• Load respective modules:

module load EIGENSOFT

module load R

module load plink

• Install the following R packages if they are not installed already:

in R:

install.packages("dplyr")

install.packages("testit")

install.packages("geepack")

• For files in eigensoft-dir folder, make sure the priviliges are set up appropriately. In eigensoft-dir:

chmod a=rwx convertf

chmod a=rwx smartpca

Execution

To submit the job, run the following shell script:

./PopCluster.sh arguments.txt

arguments.txt file contains arguments that include paths that need to be changed and default values for parameters that can be changed as well:

• WORKDIR: where the PopCluster.sh is located and where you want results to be saved.
• SCRIPTSDIR: scripts-dir folder here (contains R scripts)
• FILESDIR: files-dir folder here (contains input files (see README.md in ./files-dir for details))
• EIGENSOFTDIR: eigensoft-dir folder here (contains EIGENSOFT files for calculating PCA from genome-wide genotype data (see README.md in ./eigensoft-dir for details))
• ClusteringPCs: number of principal components used for clustering (default: 6)
• maxclustsize: minimum cluster sizet to be included in the analysis (default: 100. Make sure this number is less than or equal to number of subjects in the dataset. For the example dataset provided (mega-data-noPCs.csv in ./files-dir/) - please, change this number accordingly (< 90).)
• GLMPCs: number of principal components included in the logistic regression (default: 4)
• GLMfamily: link function to be used in the generalized linear model (default: binomial - for logistic regression)
• Nsubjects: minimum number of cases and controls in a cluster for it to be considered for the analysis (default: 5)
• MAF: minimum minor allele frequency (MAF) for a SNP in a cluster for it to be considered for the analysis (default: 0.05)
• Pvalue: significance level used for comparison of sibling leaf nodes in the pruning of a dendrogram (default: 0.05)

All the directories should be complete paths and should end in "/", for example: "/restricted/projectnb/necs/Nastia-Analysis/clustering-procedure/".

• results folder will be created after running the PopCluster.sh and will store the results of the PopCluster