Publications
Articles, preprints, and conference presentations
My published and in-progress work concerns how deep-learning models for RNA secondary structure are evaluated, and how they can be made to generalize beyond the families they were trained on. An up-to-date list is also available on Google Scholar.
Articles and Preprints
Bugnon, L. A., Kulemeyer, G., Gerard, M., Di Persia, L., Stegmayer, G., & Milone, D. H. (2026). Homology-aware cross-validation strategies for generalization assessment in RNA structure prediction.
The paper examines how homology between training and test sequences inflates reported accuracy in RNA structure prediction, and compares classical and deep-learning predictors under family-aware and sequence-similarity validation schemes. My contribution centers on the comparative benchmarking and its reproducible implementation.
In Preparation
The Less the Better: Improving RNA Secondary Structure Generalization with Structural-Aware Subsampling.
Structure-aware subsampling of training data retains roughly 18% of ArchiveII while preserving structural diversity, reaching performance comparable to training on the full dataset. The manuscript reports the subsampling strategy and its effect on cross-family generalization.
Conference Presentations
Enhancing Deep Learning Prediction of Base-Pairing Probability in RNA Sequences Using Structural Diversity.
Dissociative Adsorption of Methane on Pt(110)-(1×2) Using Machine Learning.