Mechanistic Modelling of Recessive Disease through Allelic Integration of Variant Effects

Interpreting variants in recessive diseases is difficult because clinical severity depends on the combined function of both alleles. Deep mutational scanning (DMS) experiments can provide functional measurements at scale, but their scores often relate nonlinearly to true biochemical activity. We developed a general method to infer enzymatic activities for thousands of variants by running two fitness assays at different expression levels and modelling the nonlinear activity–fitness relationship. These inferred activities allow computation of a biallelic pathogenicity score that captures the joint effect of two alleles. We applied this to adenylosuccinate lyase (ADSL), quantifying the effects of >8,000 coding variants in a yeast-based assay. The inferred activities separated pathogenic from benign alleles, and the biallelic score correlated strongly with biochemical measurements from patient-derived cells, outperforming existing predictors. This framework provides a broadly applicable strategy for mechanistic interpretation of variants in recessive enzymes. Overall design: A human ADSL variant library was generated using plasmid-based saturation mutagenesis and subsequently barcoded; each variant was associated with one or multiple unique DNA barcodes. Barcode–variant associations were determined by PacBio Sequel II sequencing. The barcoded library was transformed into the yTHC-ADE13 yeast strain, where doxycycline treatment can control expression of the endogenous ADE13 gene and enables measurement of functional complementation by human ADSL. Human ADSL variants were expressed from two distinct promoters to create low- and high-expression conditions. The pooled library was grown for five days under these expression regimes, both with and without doxycycline treatment, with samples collected every 12 hours. Barcode abundances (counts) from the growth assay were quantified on an Illumina NextSeq 2000 platform and used to calculate fitness scores from multiple barcodes per variant. This submission includes two types of sequencing samples: (1) PacBio CCS library-reference samples used to map each barcode to its ADSL variant, and (2) Illumina time-series samples collected during the yeast growth assay. Functional effects of ADSL variants were measured under four conditions reflecting the major variables under investigation: low ADSL expression with ADE13 repressed (dox-treated), low expression with ADE13 active (dox-untreated), high expression with ADE13 repressed (dox-treated), and high expression with ADE13 active (dox-untreated).