SPiP

SPiP predicts the effects of exonic and intronic variants on pre-mRNA splicing to assess their potential impact on mRNA processing and variant pathogenicity.

Key Features:

• Comprehensive analysis: Employs machine learning to perform an integrated assessment of variant impacts across 5' and 3' splice sites, branch sites, and other splicing regulatory elements.
• Curated dataset: Developed using a curated set of 4,616 variants distributed along the sequences of 227 genes, each accompanied by corresponding splicing studies.
• Bayesian analysis: Utilizes Bayesian analysis to identify control variants that lack splicing impact and to model the distribution of variants encountered in high-throughput sequencing data.
• Performance metrics: Reports 83.13% sensitivity, 99% specificity, and an AUC of 0.986, compared with AUCs of 0.965 for SpliceAI and 0.766 for SQUIRLS.

Scientific Applications:

• Genomic medicine: Supports clinical variant interpretation by predicting spliceogenicity to inform assessments of variant pathogenicity.
• Mechanistic research: Enables investigation of how nucleotide variants disrupt or create splicing motifs to elucidate molecular mechanisms underlying genetic disorders.

Methodology:

Applies machine learning techniques to predict spliceogenic effects across splice sites, branch sites, and regulatory elements, and employs Bayesian analysis to determine control variants and simulate variant distributions observed in high-throughput sequencing.