Skip to main content
Fig. 1 | Genome Biology

Fig. 1

From: GRIPT: a novel case-control analysis method for Mendelian disease gene discovery

Fig. 1

The logic flowchart of GRIPT. First, the samples of the case and control cohorts will be collected and be subjected to NGS, e.g., WES. After variant calling, the known common and/or benign variants will be filtered out based on the variant annotation and their allele frequency in large databases of normal populations and internal databases. Thus, for each gene, only a few rare variants will be left. Then, GRIPT will annotate and rank the deleteriousness of each variant, e.g., using CADD score. Based on the variant scores, a gene score will be calculated to measure the deleterious mutation load of each gene in every individual according to a given inheritance model (see the “Methods” section). Next, a Fisher’s test built upon the combination of a binomial test and a Wilcoxon rank sum test (WRST) will be calculated to measure the difference of gene score distributions between patient cohort and control cohort for each gene, and a significant p value associated with the test statistic will be assigned. This composite test is especially well suited to measure the difference of two highly skewed distributions with excesses of 0, such as the gene score distribution in the patient/control cohort computed by GRIPT (Fig. 2). Finally, according to the test statistic of each gene, GRIPT compares and ranks all genes

Back to article page