The SeqSplice multiplexed minigene splicing assay for characterization and quantitation of variant-induced and splice isoforms.

and germline variant classification is vital for clinical management of families with hereditary breast and ovarian cancer. However, clinical classification of rare variants outside of the splice donor/acceptor ±1,2-dinucleotides remains challenging, particularly for variants that induce new or cryptic splice site usage. Here, we present SeqSplice a high-throughput RNA splicing methodology utilizing barcoded minigene constructs together with a bespoke bioinformatics pipeline for identifying and quantifying the impacts for splice-altering variants. SeqSplice exhibits excellent reproducibility across cDNA input and PCR cycle differences and is able to identify and quantitate transcripts that differed by a single base. Of the 193 and 72 variants profiled, 89% (237/265) had no publicly available RNA splicing data. Complete or near complete impact owing to splice site gain/loss is observed for 42 variants, with 30 (71%) producing alternative transcripts owing to new or cryptic splice sites. These findings are used to update our aberration type predictor called SpliceAI-10k calculator, resulting in 94% specificity and 90% sensitivity for major alternative transcripts (>50% proportion). Comparison of SeqSplice findings for 28 variants with published data shows the value and limitations of using construct-based results for variant classification. Overall, our findings inform use of construct-derived data for clinical variant classification. We show that construct-derived results for variants showing low or no splicing impact provide reliable evidence against variant pathogenicity, whereas-for variants demonstrating splicing impact-construct design and naturally occurring alternative splicing are important considerations for assigning and weighting evidence towards pathogenicity.