Li Su

This paper introduces scPEFT, a parameter-efficient fine-tuning framework for scLLMs that improves task adaptation with reduced computational costs and enhances performance in disease-specific and cross-species analyses.

This study uses TRAP-seq to analyze phloem-specific translational dynamics during soybean nodulation, identifying key regulators and pathways involved in the process.

The paper presents an enhancer AAV toolbox for targeting and perturbing striatal cell types and circuits in mice, with validated specificity and cross-species applicability.

The paper develops a toolkit of transgenic mouse lines and enhancer AAVs for selective targeting of cortical cell populations using transcriptomic and epigenomic data.

scBSP identifies spatially variable genes from spatial transcriptomic data efficiently and accurately across different sequencing techniques and resolutions.

This paper creates a high-resolution transcriptomic and spatial cell type atlas of the whole adult mouse brain, revealing spatial organization and transcription factor codes for cell types.

This paper introduces BSP, a method for identifying spatially variable genes from 2D or 3D spatial transcriptomics data, showing improved accuracy and efficiency in simulations and biological validations.

RESEPT uses deep learning to visualize and characterize tissue architecture from spatial transcriptomics data, revealing pathological structures in Alzheimer's and glioblastoma samples.

DeepMAPS infers biological networks from single-cell multi-omics data using a heterogeneous graph transformer, outperforming existing tools in cell clustering and pathway prediction.