PepGM

PepGM infers taxonomic assignments from proteomic data using a probabilistic graphical model to achieve strain-level resolution for virological analyses.

Key Features:

• Probabilistic Graphical Model: PepGM employs a probabilistic graphical model to infer taxonomy from proteomic data, enabling discrimination of closely related sequences such as severe acute respiratory syndrome-related coronavirus-2 (SARS-CoV-2) strains.
• Strain-Level Resolution: Provides strain-level resolution in taxonomic assignments to distinguish closely related viral strains.
• Confidence Scores: Calculates confidence scores via belief propagation to obtain marginal distributions of taxonomic assignments and indicate when only species-level identification is supported.
• Integration with Proteomic Database Search Algorithms: Integrates results from standard proteomic database search algorithms with its probabilistic model to improve taxonomic inference accuracy.
• Performance Validation: Validated on several publicly available virus proteomic datasets, demonstrating strain-level resolution in most tested cases and reporting lower confidence when assignments are limited to species-level.

Scientific Applications:

• Virology research: Supports taxonomic inference in virology research involving rapidly evolving viruses.
• Epidemiological studies: Enables differentiation of strains to inform epidemiological analyses.
• Vaccine development: Provides strain-level information relevant to vaccine antigen selection.
• Antiviral drug design: Informs antiviral drug design by identifying strain-specific proteomic signatures.
• Viral mutation surveillance: Handles frequent viral mutations to support surveillance of rapidly evolving viruses.

Methodology:

Applies a probabilistic graphical model and belief propagation to compute marginal distributions for taxonomic assignments, integrates results from standard proteomic database search algorithms, and was validated on publicly available virus proteomic datasets.