Erik D. Huckvale

The paper trains and evaluates machine learning models to predict pathway associations of metabolites using MetaCyc data, achieving performance comparable to KEGG-based models.

The paper improves pathway enrichment analysis of metabolomics datasets by generating high-quality predicted metabolite-pathway annotations, increasing the number of detected enriched pathways by 8-fold.

The paper trains a model to predict compound pathway involvement across KEGG, Reactome, and MetaCyc using standardized chemical representations, achieving high accuracy.

The paper predicts pathway involvement of unannotated compounds using Reactome data, achieving higher accuracy than KEGG-based models.

The paper predicts pathway involvement for all KEGG pathway and compound entries using machine learning, achieving higher performance than metabolic pathways alone.

The paper introduces a machine learning model that predicts metabolite associations with KEGG Level 3 pathways, achieving high accuracy and transfer learning to Level 2 categories.

The paper predicts metabolite pathway involvement using a single binary classifier with combined metabolite and pathway features, outperforming previous methods.

The paper evaluates an erroneous KEGG-derived dataset used in machine learning for predicting metabolite-to-pathway mappings and shows how duplicates inflated model performance.

The paper develops a new benchmark dataset for predicting metabolite pathway involvement using machine learning models and evaluates their performance.

kegg_pull provides a Python package with API and CLI for efficient and flexible retrieval of KEGG database entries.