Projects

Pioneer Labs

Fitness prediction for horizontal gene transfer (2025 – present)

Using protein language models to predict how extremophile proteins will function when transferred into E. coli hosts. Benchmarking ESM2 and Profluent embeddings to understand which best captures novel biology. Developed a Bayesian probabilistic linear mixed model to infer gene fitness effects from noisy replicate data.

Columbia PhD — RNA splicing methods

LeafletFA — Atlas-scale Bayesian factor model for alternative splicing in single cells. Applied to multisample Smart-seq2 datasets to identify conserved age-associated splicing programs across mouse and human tissues. Manuscript under review at Nature Aging (2025).

SpliceVI — Variational autoencoder for joint modeling of single-cell splicing and gene expression, enabling robust integration of sparse splicing data. Pre-print: bioRxiv, 2025.

LeafletSC — Python package for alternative splicing analysis in single-cell RNA-seq using binomial mixture models.

Isoviz — R package for visualizing splice junctions across transcript isoforms, with built-in support for designing gRNA sequences targeting junctions of interest using CRISPR/Cas13.

Orthrus — Contributed to an evolutionary and functional RNA foundation model. Accepted at Nature Methods, 2026.

Columbia PhD — Course projects

Predicting COVID-19 RNA-protein interactions github.com/msthilaire5/ecbm-e4060-covid-interactions

Protein sequence-based pathogen assembly from RNA-seq github.com/karini925/VORF

Lymphoma research — UHN Princess Margaret

Worked in Dr. Robert Kridel’s lab at UHN, Toronto (May 2019 – July 2021).

NHL-ASCT-PI — Shiny web app for patient risk modeling in non-Hodgkin lymphoma, integrating clinical factors and pretransplant PET imaging. Published in Blood Advances, 2020.

Contributed to multi-omic integration studies in diffuse large B-cell lymphoma combining RNA-seq, ChIP-seq, and CRISPR/Cas9 screening (Clinical Cancer Research, 2021) and longitudinal tumor evolution analyses (Haematologica, 2021; 2022).