Oisin Mac Aodha

Large language models possess some ecologicalknowledge, but how much?

Benchmarking GPT-4o and Gemini 1.5 Pro across five closed-book ecological tasks shows solid coarse species presence and trait recall but poor fine-scale range mapping and threat classification.

Sustainable cattle management by communities supports African wildlife

The paper evaluates a community-based conservation model in Kenya, finding that cattle grazing supports wildlife but requires management to control threats like sheep/goat grazing and infrastructure proximity.

Accurate prediction of gene deletion phenotypes with Flux Cone Learning

This paper introduces Flux Cone Learning, a framework for predicting gene deletion phenotypes, showing improved accuracy over existing methods in various organisms.

Modelling dynamic host-pathway interactions at the genome scale with machine learning

The paper develops a method integrating kinetic pathway models with genome-scale metabolic models to simulate dynamic host-pathway interactions in Escherichia coli, enabling faster predictions of metabolite dynamics under genetic perturbations.

Bayesian optimization for design of multiscale biological circuits

This paper introduces a Bayesian optimization method for designing multiscale biological circuits, demonstrating its effectiveness on gene circuits with nonlinearities and multiple scales.

Accuracy and data efficiency in deep learning models of protein expression

This study compares model accuracy and data efficiency across machine learning models for predicting protein expression from DNA sequences using experimental data from E. coli and S. cerevisiae.

VideoTagger: User-Friendly Software for Annotating Video Experiments of Any Duration

VideoTagger enables efficient annotation of long video experiments, demonstrating its utility in analyzing Drosophila aging by identifying age-related mobility changes.