EpiMut

EpiMut annotates amino acid substitutions (AASs) in epigenetic factors using an alignment-free, sequence-based approach to predict their functional impact relevant to cancer and hematologic malignancies.

Key Features:

• Alignment-free sequence analysis: Operates without sequence alignments or homology searches to evaluate AASs.
• Biophysical and biochemical indices: Leverages biophysical and biochemical amino acid indices for sequence representation.
• Digital signal processing (Fourier Transform): Applies digital signal processing techniques, specifically the Fourier Transform, for sequence transformation.
• Two-step feature generation: Implements a two-step feature generation process to capture essential sequence characteristics without alignments.
• Naïve Bayes classifier: Uses a Naïve Bayes machine learning algorithm to construct predictive models that classify AASs as neutral or disease-related.
• Gene-specific models for epigenetic regulators: Builds predictive models tailored to genes encoding epigenetic regulators.
• Performance on non-conserved regions: Demonstrates superior performance for variants located outside conserved functional domains compared with PolyPhen-2, SIFT, and SNAP2.

Scientific Applications:

• Functional annotation of AASs in epigenetic factors: Provides functional impact predictions for amino acid substitutions in genes encoding epigenetic regulators.
• Identification of cancer-related variants: Enables detection and characterization of variants associated with cancer, particularly hematologic malignancies.
• Biomarker discovery and variant prioritization: Supports prioritization of variants and discovery of potential clinical biomarkers, including variants outside conserved domains.

Methodology:

Sequence-based alignment-free analysis using biophysical and biochemical amino acid indices, digital signal processing with the Fourier Transform for sequence transformation, a two-step feature generation process, and Naïve Bayes model construction.