Thibaut, L. | bioRxiv Science

bioRxiv2025

Benchmarking of variant pathogenicity prediction methods using a population genetics approach

MotivationVariant pathogenicity predictors are essential for identifying new associations between genetic variants and rare diseases. However, despite the availability of numerous predictors, there is no clear consensus on which methods provide the most reliable results. The common practice of training, testing, and benchmarking these predictors using known variant sets from disease or mutagenesis studies raises concerns about ascertainment bias and data circularity. ResultsWe benchmarked commonly used pathogenicity predictors using an orthogonal approach that does not rely on predefined "ground truth" datasets. By leveraging population-level genomic data from gnomAD and the Context-Adjusted Proportion of Singletons (CAPS) metric, we identified CADD and REVEL as the best-performing predictors for distinguishing extremely deleterious variants from moderately deleterious ones. REVEL demonstrated superior calibration. Additionally, we show that CAPS can serve as a meta-analysis tool for interpreting variant annotations and highlight biases in ClinVar-based predictor training. Availability and ImplementationCAPS analysis and benchmarking results are available at https://github.com/mgudVCCRI/PopGenVariantFiltering Contacte.giannoulatou@victorchang.edu.au

genomics↗

bioRxiv2024

Improving estimates of negative selection in human genome using CAPS

Despite ongoing efforts, variant interpretation in disease sequencing studies is often hindered by the lack of well-established ways of determining the potential pathogenicity of genetic variation, especially for understudied classes of single-nucleotide variants (SNVs). Population genetics methods offer an attractive solution to this problem by enabling the assessment of the effects of SNVs through their distributions in human populations. For instance, negative selection is known to shift site-frequency spectra of genetic variation, thus affecting the ratio of singleton variants. It has been shown that the extent of negative selection can serve as a proxy for deleteriousness. An example of this approach is the Mutability-Adjusted Proportion of Singletons (MAPS) metric. Although MAPS proves a useful instrument for the assessment of selection-based deleteriousness in SNVs, it is highly sensitive to the calibration of the singletons-by-mutability model, which results in potentially biased estimates for some classes of variants. Building up on the methodology used in MAPS, we developed a novel metric of negative selection in the human genome -- CAPS, or Context-Adjusted Proportion of Singletons. Compared to its predecessor, CAPS provides estimates of negative selection that are less biased and have more accurate confidence intervals. CAPS inherits some of the same features that make MAPS useful for studying SNVs, yet the key difference of our method is the complete elimination of the mutability layer in the model, which makes the metric more robust and reliable. We believe that CAPS holds promise for improving the discovery of new disease-variant associations in clinical and research settings.

genomics↗