Combined network analysis and machine learning allows the prediction of metabolic pathways from tomato metabolomics data
Problem
Detecting metabolic pathways traditionally relied on gene annotation; prediction from metabolomics data was needed.
Why it matters
Metabolic pathway identification is a key aspect of crop improvement and drug design.
What the AI did
A machine-learning model trained on network features of metabolite subgraphs predicted the presence of specific metabolic pathways in tomato.
Results
Predicted β-alanine-degradation-I, tryptophan-degradation-VII, β-alanine-biosynthesis-III, and melibiose-degradation pathways; in vivo assays mentioned but abstract truncated
Impact
Predicted candidate metabolic pathways relevant to crop improvement and drug design.
Record
| Venue | Communications Biology |
|---|---|
| Published | 2019 |
| Work type | article |
| Authors | David Toubiana, Rami Puzis, Lingling Wen, Noga Sikron, Assylay Kurmanbayeva, Aigerim Soltabayeva, María del Mar Rubio Wilhelmi, Nir Sade et al. |
| Citations | 87 (OpenAlex, at import time) |
| References | 54 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s42003-019-0440-4 |
| Open access | gold |
| Keywords | metabolomics, machine learning, metabolic pathways, tomato, network analysis |
Importance rating
real but narrower advance
Abstract calls pathway identification a key aspect of crop improvement and drug design; demonstrated only on tomato pathways; no scale stated.