I have performed targeted sequencing on a panel of genes resulting in variant data for a cohort consisting of patients with different cancer diagnoses. These same genes have been analyzed in other cohorts and we wish to compare the prevalence of variants across the different cohorts.
While one can compare the overall number of variants as a total % of the cohort (ex. Cohort A has variants in 5% of cases and Cohort B has variants in 10%), this becomes complicated when the cohorts are comprised of different distributions of cancer types (ex. Cohort A has 20% Leukemia patients and Cohort B has 40% Leukemia patients - and Leukemias are known to have a higher percentage of mutations compared to other cancers).
I have tried to adjust the distribution of the cancer subtypes in my cohort to match the distribution of the individual subtypes in the comparative cohorts and then applying adjustment factors to the number of mutations detected in each subtype in my cohort - therefore producing an estimation of the number of mutations I would have detected (based on our detected variants) had our cohort had the same distribution as the one we are comparing to. However this also raises some biases and comparisons that I do not statistically know how to address: for instance if we have detected variants in a subtype of cancer in our dataset which is not included in the subtypes of the comparative cohort, then our number of observed variants would be adjusted to zero based on the distribution adjustment approach.
I was wondering if someone had a suggestion on how I could go about this in a better way?