IMN4NPD: An Integrated Molecular Networking Workflow for Natural Product Dereplication

Area of Science:

• Natural Product Discovery
• Computational Chemistry
• Bioinformatics

Background:

• Molecular networking is a standard tool for natural product (NP) discovery.
• Current methods often overlook smaller clusters, such as self-looped or paired nodes, in molecular networks.

Purpose of the Study:

• To develop an improved workflow for NP dereplication that addresses limitations of current methods.
• To enhance the identification and annotation of both large clusters and overlooked smaller nodes in molecular networks.

Main Methods:

• Developed the integrated molecular networking workflow for NP dereplication (IMN4NPD).
• Integrated outputs from multiple computational tools for compound dereplication and annotation.
• Utilized distinct spectral similarity algorithms (Spec2Vec, MS2DeepScore) for t-SNE network construction.
• Compared performance against modified cosine similarity.

Main Results:

• IMN4NPD expedites dereplication of extensive clusters and prioritizes self-looped or paired nodes.
• Integration of diverse spectral similarity measures improved NP dereplication capabilities.
• Demonstrated utility with an ethanol extract of *Plumula nelumbinis*, aiding dereplication and novel compound discovery.

Conclusions:

• IMN4NPD offers a more comprehensive approach to NP dereplication by including all node types.
• The integration of multiple computational solutions and diverse similarity algorithms significantly enhances NP discovery.