It is challenging to infer taxonomy in shotgun proteomics studies using mass spectrometry. Protein presence and matching taxa must be inferred from a list of detected peptides in multi-species or viral samples of unknown taxonomic origin, which is sometimes made difficult by protein homology: many proteins share peptides not just within a taxon but also between taxa.
The authors introduce PepGM, a probabilistic graphical model with strain-level resolution and associated confidence scores for the taxonomic assignment of virus proteomic samples. To determine the marginal distributions, and consequently confidence scores, for potential taxonomic designations, PepGM combines the outcomes of a common proteomic database search algorithm with belief propagation. Only species-level taxonomic designations were valid in two of eight situations, as shown by PepGM’s lower confidence scores.
PepGM is available at https://github.com/BAMeScience/PepGM.
Reference:
Holstein T.(2023)PepGM: a probabilistic graphical model for taxonomic inference of viral proteome samples with associated confidence scores 39(5):btad289