Fuyi Li

CellReg3D integrates DNA language models and 3D genome structure to model cell-type-specific gene regulation, achieving state-of-the-art performance in predicting epigenetic marks and gene expression.

DesignMaster improves PROTAC design by incorporating linker length and physicochemical constraints through a diffusion framework, achieving better validity and recovery than existing methods.

MolX is a geometric foundation model that jointly learns representations of protein pockets and ligands from 3D structural data, achieving state-of-the-art performance in predicting molecular interactions.

hoodscanR identifies and analyzes cell neighborhoods in spatial transcriptomics data, revealing transcriptional differences in tumor cells across neighborhoods.

DOMINO identifies spatial domains in spatial transcriptomic data with improved accuracy and scalability, outperforming existing methods on large datasets.

GatorDuo improves spatial domain identification in spatial transcriptomics by refining dual graphs with pseudo-label agreement and adaptive clustering.

GatorSC improves single-cell transcriptomic analysis by integrating multi-scale cell and gene graphs with a Mixture-of-Experts fusion framework, outperforming existing methods in clustering, imputation, and annotation.

This study introduces Termini, an AI framework for designing antimicrobial peptides with terminal modifications, achieving high hit rates and therapeutic potential in vitro and in vivo.

OmniCleave models protease-substrate specificity using a geometric graph framework, outperforming existing methods and identifying novel substrates.

GRASP uses a graph neural network to predict RNA subcellular localization by modeling RNA structure, outperforming existing methods across RNA types and transcript lengths.

This paper introduces Kinhibit, a framework that uses pre-trained molecular encoders and protein language models to predict kinase-inhibitor binding affinity, achieving high accuracy on MAPK pathway kinases and a large kinase dataset.

Diet-Seg improves brain tumor segmentation in MRI by using entropy-based hardness estimation and dynamic learning rates, outperforming existing methods across multiple datasets.

DegradeMaster predicts PROTAC-targeted protein degradation using a semi-supervised E(3)-equivariant graph neural network, achieving improved accuracy over existing methods.

MetalPrognosis predicts disease-associated mutations in metal-binding sites using a biological language model and convolutional neural network, outperforming existing methods.

This paper introduces scSimGCL, a graph contrastive learning framework for scRNA-seq data that improves cell clustering through self-supervised pretraining and high-quality representation learning.

The paper introduces Mimosa, a Transformer-based model for miRNA target prediction that incorporates non-canonical base-pairing patterns, achieving better performance than traditional methods.

The paper introduces Allocator, a deep learning framework for predicting mRNA subcellular localization by integrating sequence and structure information, showing improved performance over existing methods.

MERITS provides a comprehensive database of PE/PPE proteins from Mycobacterium, offering structural and functional information to aid in vaccine and diagnostic development.

The paper develops a deep learning model to predict protein degradation by proteases and UVR/ROS in human skin, validated against experimental data and integrated into a web tool for proteome susceptibility analysis.

PRISMOID is a 3D structure database for post-translational modifications and mutations, providing structural and functional insights into PTM sites and their interactions.