PLANTMETBENCH: A MULTIMODAL KNOWLEDGE GRAPH BENCHMARK FOR BIOSYNTHESIS PREDICTION

Abstract

Plant metabolism produces a vast diversity of molecules, yet identifying the enzymes that catalyse biosynthetic reactions remains a major open problem. Existing benchmarks typically reduce this problem to isolated enzyme–reaction pairs, overlooking pathway and metabolite context that determines biological plausibility, leaving no standardised way to evaluate approaches that exploit this richer multimodal and relational structure. We present PlantMetBench, a multimodal graph benchmark built on PlantMetWiki, a curated knowledge graph of plant metabolism spanning 424 organisms. The resulting benchmark graph comprises 13,682 proteins, 2,741 genes, 4,577 metabolites, and 11,980 experimentally curated enzyme–reaction pairs. PlantMetBench frames enzyme–reaction prediction as an organism-disjoint link-prediction task and provides graph-native dataset splits, precomputed multi-modal node features, and evaluation protocols. Using PlantMetBench, we evaluate heterogeneous graph neural networks, standard architectures (HeteroSAGE, GAT, HGT, and R-GCN), alongside residual and jumping-knowledge variants, against non-graph models. Most heterogeneous graph architectures consistently outperform non-graph baselines on our rank-based evaluation, suggesting the value of relational context for enzyme–reaction prediction. Yet, performance remains far from saturated, indicating substantial room for improved representations and modelling strategies. Also, we show that benchmark design choices (shortcuts, splits, and metrics) are key to aligning evaluation with biologically meaningful tasks. PlantMetBench provides a standardised, reproducible framework to evaluate multimodal biosynthetic data, establishing a foundation for future work on graph-based architectures for enzyme–reaction prediction. Code and data are publicly available.

DICD-IE: an expert-annotated benchmark for evaluating pretrained biomedical language models in dental implant literature and clinical knowledge extraction
Almonzer Salah Nooraldaim, Azza A. Abushama, Hiba Ahmed +7
Scientific Reports30 Sep 2026Peer-reviewedCC BY-NC-ND 4.0Full textMolecular biology & gene regulationCrossref
Explainable/interpretable artificial intelligence and knowledge graph neural network for synthetic lethality prediction: overview, challenges, and future directions
Briefings in bioinformatics1 Sep 2026Peer-reviewedCC BY-NCFull textMolecular biology & gene regulationEurope PMC
Pep2Net: an integrated peptide knowledge base and cross-attention model for multi-task prediction of peptides
Boyan Zheng, Jiahao Yuan, Yuzhou Wang +4
Genome Biology14 Oct 2026Peer-reviewedCC BY-NC-ND 4.0Molecular biology & gene regulationCrossref
SaltyMeta: a curated benchmark and protein language model-informed web tool for salty peptide prediction
Wanchao Chen, Wen Li, Yanan He +1
arXiv15 Sep 2026PreprintCC BY-SA 4.0Full textBiophysicsarXiv
Joint enzyme–reaction retrieval and catalytic optima prediction via multimodal fusion
Yueyi Cai, Feng Yang, Juan Liu
Bioinformatics12 Oct 2026Peer-reviewedCC BY 4.0Full textMolecular biology & gene regulationCrossrefPMC

Cite

BibTeX RIS

BibTeX

@misc{pup2026plantmetbench,
  title = {{PLANTMETBENCH: A MULTIMODAL KNOWLEDGE GRAPH BENCHMARK FOR BIOSYNTHESIS PREDICTION}},
  author = {Elena Del Pup and Berfin Inal and Joana Tirana and Jose Eduardo Escrig Molina and Yu-Long Xu and Aike Potze and Justin J.J. van der Hooft and Marnix H. Medema and Daniel Probst},
  howpublished = {ChemRxiv},
  year = {2026},
  month = oct,
  doi = {10.26434/chemrxiv.15011165/v1},
  url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15011165/v1},
  note = {CC BY},
}

RIS

TY  - UNPB
TI  - PLANTMETBENCH: A MULTIMODAL KNOWLEDGE GRAPH BENCHMARK FOR BIOSYNTHESIS PREDICTION
AU  - Elena Del Pup
AU  - Berfin Inal
AU  - Joana Tirana
AU  - Jose Eduardo Escrig Molina
AU  - Yu-Long Xu
AU  - Aike Potze
AU  - Justin J.J. van der Hooft
AU  - Marnix H. Medema
AU  - Daniel Probst
PY  - 2026
DA  - 2026/10/27
PB  - ChemRxiv
DO  - 10.26434/chemrxiv.15011165/v1
UR  - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15011165/v1
AB  - Plant metabolism produces a vast diversity of molecules, yet identifying the enzymes that catalyse biosynthetic reactions remains a major open problem. Existing benchmarks typically reduce this problem to isolated enzyme–reaction pairs, overlooking pathway and metabolite context that determines biological plausibility, leaving no standardised way to evaluate approaches that exploit this richer multimodal and relational structure. We present PlantMetBench, a multimodal graph benchmark built on PlantMetWiki, a curated knowledge graph of plant metabolism spanning 424 organisms. The resulting benchmark graph comprises 13,682 proteins, 2,741 genes, 4,577 metabolites, and 11,980 experimentally curated enzyme–reaction pairs. PlantMetBench frames enzyme–reaction prediction as an organism-disjoint link-prediction task and provides graph-native dataset splits, precomputed multi-modal node features, and evaluation protocols. Using PlantMetBench, we evaluate heterogeneous graph neural networks, standard architectures (HeteroSAGE, GAT, HGT, and R-GCN), alongside residual and jumping-knowledge variants, against non-graph models. Most heterogeneous graph architectures consistently outperform non-graph baselines on our rank-based evaluation, suggesting the value of relational context for enzyme–reaction prediction. Yet, performance remains far from saturated, indicating substantial room for improved representations and modelling strategies. Also, we show that benchmark design choices (shortcuts, splits, and metrics) are key to aligning evaluation with biologically meaningful tasks. PlantMetBench provides a standardised, reproducible framework to evaluate multimodal biosynthetic data, establishing a foundation for future work on graph-based architectures for enzyme–reaction prediction. Code and data are publicly available.
ER  -

Abstract

Related

Cite