9 CHAPTER 9 198 variants were used in the analysis. For the gnomAD control dataset, only summary statistics were available. Therefore, to approximate the number of control subjects carrying at least one qualifying variant in a given gene, the allele counts for all qualifying variants in that gene were summed. This summation based approximation probably is an overestimation as it is likely that some individuals carry multiple variants in the same gene. Contrary to rare variant analysis where only the locations of the qualifying variant in cases are used for controls, we selected all variants across the entire gene in controls, in the same way as what was done for the cases. As a result, we had a total number of all missense variants per CACNA1x gene in both cases and controls. These “qualifying variants” for both the case and the control cohort were compared. Insertions and deletions (Indels) were not included in the analysis due to their higher percentage of sequencing artefacts, especially given the differing sequencing platforms used across cohorts. Multiple-variant burden testing of CACNA1x genes. Gene-based burden testing was performed for all variants that met the quality filters, which are referred to as “qualifying variants”, using adaptation on the TRAPD test (Testing Rare vAriants using Public Data).39 TRAPD was chosen because the control dataset consisted of summary data rather than individual-level genotype data as well as for its approach to collate variants which mitigates the statistical drawbacks of burden testing per variant or per individual. The TRAPD test was implemented to determine whether CACNA1x genes and subjects carried a significant burden of missense variants in cases. TRAPD produces counts of “collapsed” variant groups across each gene and for the respective case or control cohort. To conduct the test, a group file with the qualifying variants was created for each of the eight genes (CACNA1A, CACNA1B, CACNA1C, CACNA1D, CACNA1E, CACNA1G, CACNA1H and CACNA1I). Of note, CACNA1S was excluded from the analysis as it encodes the pore-forming CaV1.1 α1S subunit that is exclusively expressed in skeletal muscle, so not in the brain and CACNA1F was excluded as this gene is located on the X-chromosome and TRAPD is currently not configured to test non-autosomal chromosomes. We performed gene-based burden testing for all single-point variants in each cohort. The following steps, in brief, were performed: (1) variants for each CACNA1x gene in the case group were compiled into a “SNP file”, 2) a python script was used to interrogate the VCFs and count the occurrence of variants in each gene in both the case and the controls cohorts independently. This generated variant count data for each gene and (3) the one sided Fisher’s exact test was used on the allele count tables to identify the probability of excess in the number of allele counts in cases relative to controls (i.e. the statistical significance of the burden). (4) the one sided Fisher’s exact test was used on the subject count tables to identify the probability of excess in the number of subjects with variants in cases relative to controls (i.e. the statistical significance of the burden). P-values < 6.25 x 10-3 were considered significant (Bonferroni corrected for testing 8 genes). Odds ratios were calculated to assess the magnitude of the burden effect. Genes exhibiting statistically significant burden in HM from the Australian discovery cohort were also tested in the Dutch replication cohort.
RkJQdWJsaXNoZXIy MTk4NDMw