Yingzhou Lu

BiomniBench evaluates LLM agents' research processes in biomedical tasks using expert rubrics, revealing performance gaps in method selection and reasoning.

The paper examines how diet affects gut microbiome and plasma metabolome across insulin resistance states, finding stronger diet-omics associations in insulin-sensitive individuals.

An LLM agent with a literature-mined environment of 150 tools, 105 packages and 59 databases plans in code, matching reported human accuracy on database and sequence question answering.

ABDS provides tools for analyzing biologically diverse samples by addressing missing value imputation, signature gene detection, and expression pattern visualization.

DDN2.0 provides R and Python tools for differential dependency network analysis to identify key pathways and network rewiring across omics data.

This paper introduces OVE-FC and OVE-sFC tests to detect subtype-specific and differentially expressed genes, showing improved accuracy in deconvolution compared to existing methods.

COT detects marker genes across subtypes using cosine similarity in a one-sample test, applicable to multi-omics data.

This paper proposes two computational methods to integrate multi-omics data and identify disease-specific pathways, demonstrating their effectiveness on muscular dystrophy data.