Fig. 1

Overview of the steps required to produce, validate and use multiplexed functional data for variant interpretation. Multiplexed assays for variant effect. a A DNA variant library is generated and introduced into cells before being b subjected to a functional assay. c Variants from a sample of each cell population are sequenced and d functional scores that reflect their change in frequency are calculated for each variant. Data quality control. e The dynamic range of the functional score distribution of the entire library (gray) is benchmarked by the observed scores for known functionally normal (blue) or abnormal (red) variants; here, synonymous and nonsense protein variants are used as an example. f. Comparison of functional scores across two or more replicate experiments generates an overall metric of reproducibility (R²). Biological replicates have different input populations that result from separate introductions of the variant library and they provide a better characterization of variation than technical replicates, which use the same starting population. g Confidence intervals for each variant functional score are calculated from replicate experiments. h Multiplexed functional scores are benchmarked against other measurements of molecular function. Reporting. i Sharing of data and analyses facilitates data reuse and enhances data utility. Validation for clinical utility. j Functional score ranges are divided into categorical bins. Here, the cutoff between bins is determined by the measured functional score distributions of known interpreted variants. k A precision-recall curve is used to assess the assay’s sensitivity and specificity. Multiplexed functional data as evidence for variant interpretation. l Clinicians and diagnostic laboratories assess the overall quality of the assay and of specific variant information to determine the weighting, in terms of strength of evidence that should be assigned to the functional information