Bo Li

The paper evaluates whether label-free morphology can replace targeted cellular measurements by analyzing paired phase-reporter data across gene knockouts and screens.

MVCBench evaluates drug molecular and gene representation methods for predicting drug-induced multimodal cell phenotypes using transcriptomic and imaging data.

SpaIM improves spatial transcriptomics by imputing missing gene expressions using scRNA-seq data, outperforming 12 existing methods across 53 datasets.

This paper introduces HGGEP, a hypergraph neural network model that predicts gene expression from histology images by capturing cell morphology and higher-order feature associations.