Lipeng Lai

siDiff: De Novo siRNA Design via Efficacy-Guided Discrete Masked Diffusion

siDiff improves siRNA design by combining a diffusion model with an efficacy guidance model, achieving better hit rates and generalization across target genes.

OCOO-T : A SIMPLE AND SCALABLE VIRTUAL CELL MODEL FOR TRANSCRIPTIONAL PERTURBATION RESPONSE PREDICTION

OCOO-T predicts transcriptional responses to perturbations using a minimalist Transformer-based model, achieving state-of-the-art performance across cell types and perturbations.

BioScientist Agent: Designing LLM-Biomedical Agents with KG-Augmented RL Reasoning Modules for Drug Repurposing and Mechanistic of Action Elucidation

The paper introduces BioScientist Agent, a framework combining a biomedical knowledge graph, variational graph auto-encoder, reinforcement learning, and LLMs to improve drug repurposing and mechanistic reasoning.

Pre-training with a rational approach for antibody sequence representation

This study introduces PARA, a pre-trained model for antibody sequences that outperforms existing models on various tasks by incorporating antibody-specific features.

SC-MAMBA2: Leveraging State-Space Models for Efficient Single-Cell Ultra-Long Transcriptome Modeling

SC-MAMBA2 is a single-cell foundation model using state-space models to efficiently process ultra-long transcriptome sequences, outperforming existing models in accuracy and efficiency.

WUREN: Whole-modal fUsion Representation for protein interaction interfacE predictioN

WUREN is a multimodal model that fuses sequence, graph, and structural features to predict protein interaction interfaces and antigen epitopes, achieving state-of-the-art performance.

Single point mutations can potentially enhance infectivity of SARS-CoV-2 revealed by in silico affinity maturation and SPR assay

The paper predicts single point mutations in SARS-CoV-2 RBD that enhance binding affinity to hACE2 using in silico methods and validates them with SPR assays.