Fig. 1: Overview of the method to detect NVEs and examples.
From: Widespread naturally variable human exons aid genetic interpretation
A Model figure describing NVE detection and our summary statistic, exon frequency. To define NVEs, we first filtered LeafCutter outputs for either alternative splice sites or skipped exons. For a given NVE observed in the RNA sequencing data, the Ψ of an individual can be directly estimated using the beta binomial model. The EF summary statistic is the percentage of individuals whose Ψ exceeds a given threshold, shown for a 5% Ψ threshold. B An example exon in FBXO6 with EF5% = 0.07 in heart tissue. Individuals with heart samples (n = 386) are sorted by their estimated Ψ values for this exon, which is shown as percentiles on the x-axis. The mean posterior estimate of Ψ is plotted (highlighted in purple if the estimate is at least 5% Ψ in the individual), along with error bars showing one standard deviation on Ψ. In the inset, an example of the population distribution used to calculate the exon frequency (EF) of this exon. Note that the y-axis scales differ between plots. C An example of an NVE in the CAST gene (with EF5% = 0.5) that has high variability amongst individuals. Individuals with Esophagus Mucosa samples (n = 497) are sorted by their estimated Ψ values for this exon, which is shown as percentiles on the x-axis. Inset: as in B. Below: bar plots of EJ reads for both NVE and cognate, and overall mRNA levels (TPM). Note that the y-axis scales differ between plots.
