Athan Z. Li

A sparse mixture-of-experts DNA model conditioned on hyperbolic taxonomy embeddings learns better microbial representations than its taxonomy-free twin and comes close to Evo2-7B in probing and fine-tuning at lower compute.

This study performs a single-nucleus transcriptome-wide association study to identify gene-trait associations in brain disorders, revealing cell-type-specific genetic regulation and ancestry-specific patterns.

The paper constructs a single-nucleus atlas of genetic regulation in the human brain, identifying cell-type-specific regulatory effects and disease-associated genes.

This paper benchmarks 20 computational methods for multi-omics biomarker discovery in cancer using real and simulated datasets, finding that DeePathNet and Deep-KEGG perform best.

NeuroTD infers time-varying neural activity delays using adaptive sliding windows, showing longer delays in hippocampal signals during working memory tasks and gene expression correlations in mouse motor cortex data.