Machine Learning 🚀 In Genomics 🧬 and HealTh ❤️ = 💡

Welcome to Li Lab at McGill Computer Science

Our research focuses on developing AI methods for computational biology, particularly in the areas of population genetics, single-cell multi-omics, and electronic health records (EHR). We are passionate about leveraging machine learning to make a translational impact in healthcare.

Recent Research Highlights

MixEHR

Multimodal EHR integration

Inferring multimodal latent topics from electronic health records. Nature Communications 2020

MixEHR

Federated learning in EHR

FedWeight: mitigating covariate shift of federated learning on EHR data through patients re-weighting. npj Digital Medicine 2025

TimelyGPT

Time-series health forecasting

TimelyGPT: Extrapolatable Transformer Pre-training for Long-term Time-Series Forecasting in Healthcare. Health Information Science and Systems 2025

TrajGPT

Irregular time-series health forecasting

TrajGPT: Irregular Time-Series Representation Learning of Health Trajectory. IEEE Journal of Biomedical and Health Informatics 2025

VIPRS

Bayesian polygenic risk score inference

Fast and Scalable Polygenic Risk Modeling with Variational Inference. American Journal of Human Genetics 2023

VIPRS

Ultra-efficient polygenic risk score inference

Toward whole-genome inference of polygenic scores with fast and memory-efficient algo-rithms American Journal of Human Genetics 2025

SSLPRS

Bayesian PRS inference with Spike-and-Slab LASSO prior

Sparse Polygenic Risk Score Inference with the Spike-and-Slab LASSO. Bioinformatics 2025

SparsePro

Inferring causal genetic variants

SparsePro: an efficient genome-wide fine-mapping method integrating summary statistics and functional annotations. PLOS Genetics 2023

GTM-decon

Cell-type deconvolution

Guided-topic modelling of single-cell transcriptomes enables joint cell-type-specific and disease-subtype deconvolution of bulk transcriptomes with a focus on cancer studies. Genome Biology 2023

scCello

Single-cell foundation model

Cell ontology guided transcriptome foundation model. NeurIPS 2024 spotlight.

scETM

Single-cell embedded topic model

Learning interpretable cellular and gene signature embeddings from single-cell transcriptomic data. Nature Communications 2021

scETM

Genome foundation model for scATAC-seq analysis

GFETM: Genome Foundation-based Embedded Topic Model for scATAC-seq Modeling. RECOMB 2024

Latest News