PaPI

PaPI predicts the functional impact of coding variants by encoding wild-type and mutant protein sequences with pseudo amino acid composition and scoring variants using a machine-learning classifier that integrates amphiphilic composition, evolutionary conservation, and homologous-protein information.

Key Features:

• Pseudo Amino Acid Composition: Represents wild-type and mutated protein sequences using pseudo amino acid composition to capture sequence patterns relevant to variant impact.
• Machine Learning Classifier: Employs a machine learning classifier trained on known deleterious and benign coding variants to score unobserved variants for likelihood of functional damage.
• Integration of Multiple Methods: Combines amphiphilic pseudo amino acid composition with evolutionary conservation and homologous proteins–based features to enhance predictive accuracy.
• Handling Complex Variants: Scores complex coding variants including deletions, insertions, and indels.
• High-throughput Scoring: Processes large sets of coding variants (thousands per run) for large-scale genomic studies.

Scientific Applications:

• Genetic disease diagnosis: Prioritizes coding variants for genetic disease studies by predicting their likelihood of being deleterious.
• Personalized medicine: Informs interpretation of patient-specific coding variants to support clinical decision-making.
• Large-scale variant interpretation: Enables high-throughput scoring of thousands of variants for population, cohort, and research-oriented genomic analyses.

Methodology:

Protein sequences are encoded via pseudo amino acid composition (including amphiphilic descriptors); a machine learning classifier trained on labeled deleterious and benign coding variants is used for prediction, incorporating evolutionary conservation and homologous-protein analyses.