Exploitation – CoDiet

CoDiet Outputs

The CoDiet project is comprised of 7 work packages (WPs), the interaction between these can be seen below.

WP1: Technology-assisted literature triage

https://doi.org/10.1101/2025.09.04.673740 – Publication (preprint) accompanying the CoDiet corpus
https://doi.org/10.64898/2026.03.03.709244 – Publication (preprint) accompanying the CoDiet relation extraction methods
https://doi.org/10.1093/bioinformatics/btag439 – Publication accompanying the CoDiet relation extraction methods
https://doi.org/10.1101/2025.08.29.671515 – Publication (preprint) demonstrating the concept of AI-assisted literature review for microbiome research
https://doi.org/10.1093/bioinformatics/btag418 – Publication (preprint) demonstrating the concept of AI-assisted literature review for microbiome research

https://doi.org/10.5281/zenodo.22097875 – Data of original publications and test set to be used for evaluation
https://doi.org/10.5281/zenodo.22097879 – Annotation data for publications
https://doi.org/10.5281/zenodo.22097881 – Annotated dataset that can be used for training
https://doi.org/10.5281/zenodo.19050553 – Data used to train and evaluate the model
https://doi.org/10.5281/zenodo.17305410 – Data used to train and evaluate the model
https://www.codabench.org/competitions/10913/ – Benchmark data to allow new models to be evaluated without the risk of train-test leakage

https://github.com/omicsNLP/CoDietCorpus – Code to annotate text documents using different methods for entity recognition
https://github.com/omicsNLP/CoDietZenodo – Code to recreate the CoDiet Corpora from scratch
https://github.com/omicsNLP/RECoDe – Code to use the model to perform relation extraction
https://github.com/omicsNLP/microbELP – Code to use the model to annotate microbes in text

Dr Joram Posma – Associate Professor, Department of Metabolism, Digestion and Reproduction, Imperial College London

Ms Ivana Balkan – Innovation Manager, Imperial College London