Case-Control Analysis

OpenCRAVAT can perform basic comparisons between case and control cohorts in a study. The Case/Control feature allows users to (1) count samples in case and control groups with particular genotypes, (2) compute the one-sided p-value using Fisher’s Exact Test. Case and control sets are defined by a separate input file.

Installing

Install the casecontrol module. This module may not be available in older versions of OpenCRAVAT.

oc module install casecontrol

Case/Control requires scipy to perform statistical tests. On most systems, this can be installed with pip3 install scipy, but consult the scipy website for more specific instructions.

Running

Case-Control may be run from the command line in most OpenCRAVAT versions, and from the GUI in versions above 2.2.0.

Both methods will require a text file which assigns samples to a cohort. The file contains two columns, with whitespace as the delimiter.

sample_0    case
sample_1    control

Samples which are not in the cohorts file, or are assigned a cohort other than case or control, will not be included in the analysis.

You will be notified if samples in the cohorts file cannot be found in the job. In some cases, this is because multiple input files were used. In that situation, the sample ID must be prefixed with the filename of the input it came from, and a double underscore. For example, a sample called sample_0 from file input.vcf would become input.vcf__sample_0. ### Command line

Run casecontrol in the command line by pointing the module to the cohorts file.

oc run input --module-option casecontrol.cohorts=cohorts.txt

GUI

Run casecontrol in the GUI by scrolling to the bottom of the left hand panel in the submit page, to the section called “Additional Analysis”. Then, click the “Case-Control cohorts” button to select your cohorts file.

Interpreting results

There will be nine output columns. The three columns shown by default are p-values of the likelihood that a variant occurs more in case samples under three different inheritance models. Six hidden columns include counts of homozygous, heterozygous and reference variants across the cohorts.

P-value calculation for inheritance modes

For the Dominant model, we create a 2x2 contingency table to assign a p-value using a Fisher’s exact test. The first column includes the number of samples that have any alternate allele, whether heterozygous or homozygous.

	Alt (Aa and aa)	Ref (A/A)
Cases	N11	N12
Controls	N21	N22

For the Recessive model, we create a 2x2 contingency table to assign a p-value using a Fisher’s exact test. The first column includes the number of samples that are homozygous for the alternate allele.

	Alt (aa)	Ref (AA and aa)
Cases	N11	N12
Controls	N21	N22

For the allelic model, we create a 2x2 contingency table to assign a p-value using a Fisher’s exact test. The first column includes the count of non-reference genotypes and the second column includes the count of reference genotypes.

	Alternate genotypes	Reference genotypes
Cases	N11	N12
Controls	N21	N22