Improving annotation propagation on molecular networks through random walks: introducing ChemWalker

Area of Science:

• Computational chemistry
• Bioinformatics
• Analytical chemistry

Background:

• Mass spectrometry analysis of complex mixtures is hindered by the bottleneck of mass signal annotation.
• Molecular networks are increasingly used for annotating large-scale mass spectrometry experiments.
• Network Annotation Propagation (NAP) automates annotation propagation but is limited to local information.

Purpose of the Study:

• To introduce ChemWalker, a Python library that expands annotation propagation in molecular networks.
• To overcome the limitation of local information propagation in existing methods like NAP.

Main Methods:

• ChemWalker utilizes random walks on graphs for annotation propagation.
• It employs in silico fragmentation (MetFrag) and biologically relevant databases.
• A weighted graph is created using MetFusion Scoring to guide the random walk process.

Main Results:

• Annotation propagation is expanded to nodes not directly connected to spectral matches.
• ChemWalker demonstrates improvements in running time, scalability, and maintainability compared to NAP.
• The library enables efficient analysis of complex mixtures through enhanced annotation.

Conclusions:

• ChemWalker offers a more efficient and scalable approach to spectral annotation in mass spectrometry.
• The random walk method effectively propagates information across molecular networks.
• This tool facilitates the analysis of complex biological samples by improving annotation accuracy and speed.