07_Case_control_analysis.md

Case-control analysis

The case-control analysis is performed by a Python3 script casecontrol.py located inside scripts directory. The locus-based analysis can be run like so:

python3 casecontrol.py locus \
    --manifest manifest.txt \
    --multisample-profile multisample_profile.json \
    --output-prefix output

The command to run the motif-based analysis is nearly identical:

python3 casecontrol.py motif \
    --manifest manifest.txt \
    --multisample-profile multisample_profile.json \
    --output-prefix output

The input parameters manifest.txt and multisample_profile.json are as described previously.

Optional parameter	Description	Default
--min-count	Minimum number reads in a region for downstream analysis	5
--target-regions	BED file with regions to which analysis should be restricted	NA
--test-method	Method of calculating Wilcoxon Rank-Sum Test p-value*	normal

* The default value normal invokes the Normal approximation appropriate for larger samples. To compute the p-value directly for smaller samples, use permute_<N> where N is the number of permutations. For example, permute_1000000 invokes a test with 1000000 permutations.

Outputs

The program produces two output files. One of them summarizes per-locus comparison of in-repeat reads. The other file summarizes the overall genome-wide comparison of motifs.