ChemWalker

ChemWalker propagates annotations across molecular networks using random walks on weighted graphs to increase annotation coverage in untargeted mass spectrometry analyses of complex mixtures.

Key Features:

• Random Walk-Based Annotation Propagation: Expands propagation beyond first neighbors by performing random walks on molecular networks to spread annotations from spectral matches.
• Integration with MetFrag and MetFusion: Leverages MetFrag combinatorial in silico fragmentation results and the MetFusion scoring function to inform candidate evaluation and weighting.
• Weighted Graph Construction: Combines spectral network data with structural similarities among candidate structures to transform molecular networks into weighted graphs.
• Seed-Based Propagation: Initiates random walks from seed nodes corresponding to spectral library matches to compute traversal probabilities to candidate nodes.
• Performance Improvements: Offers improved running time, scalability, and maintainability compared to Network Annotation Propagation (NAP).

Scientific Applications:

• Large-scale untargeted mass spectrometry annotation: Increases coverage and accuracy of annotations across molecular networks in untargeted MS experiments.
• Biomolecule identification and pathway analysis: Supports identification of candidate structures and interpretation of biochemical pathways via propagated annotations.

Methodology:

Uses MetFrag combinatorial in silico fragmentation results and the MetFusion scoring function to define a weight function; constructs weighted graphs by combining spectral network edges with structural similarity among candidate structures; performs random walks starting from seed nodes (spectral library matches) to compute traversal probabilities and propagate annotations beyond first neighbors.