MetaNovo: An open-source pipeline for probabilistic peptide discovery in complex metaproteomic datasets

Area of Science:

• Microbiology
• Proteomics
• Bioinformatics

Background:

• Microbiome research offers insights into metabolic interactions relevant to disease, agriculture, and climate change.
• Inferring microbial protein synthesis from RNA and protein expression data is challenging due to poor correlations.
• Current metaproteomic analyses often rely on limited, knowledge-based sequence databases, potentially missing diverse proteins.

Purpose of the Study:

• To introduce MetaNovo, a novel computational approach for metaproteomic analysis.
• To enable accurate protein identification directly from mass spectrometry data without prior knowledge of sample composition.
• To overcome limitations of existing methods relying on metagenomic or whole genome sequencing data.

Main Methods:

• MetaNovo combines open-source tools for de novo sequence tag matching.
• It employs a novel algorithm for probabilistic optimization of the UniProt knowledgebase.
• This creates tailored sequence databases for proteome-level, target-decoy searches.

Main Results:

• MetaNovo demonstrated comparable peptide and protein identifications to existing pipelines on human samples, with enhanced identification of non-bacterial peptides.
• Benchmarking on known microbial samples yielded more MS/MS identifications for expected taxa and improved taxonomic representation.
• The approach identified an experimental sample contaminant without prior expectation and highlighted genome sequencing quality concerns.

Conclusions:

• MetaNovo enables simultaneous identification of peptides from all domains of life in metaproteome samples directly from tandem mass spectrometry data.
• This method bypasses the need for curated sequence databases, offering a more accurate alternative to current gold standard approaches.
• MetaNovo can identify unexpected sample contaminants and reveal previously unidentified metaproteomic signals.