Wenze Ding

This paper benchmarks decision strategies for protein engineering, showing when models and measurements improve decisions based on accumulated evidence.

The paper reanalyzes plant stress transcriptomes to distinguish biological corroboration from analytical depth using a claim-centred atlas called PLEASE.

BRIDGE-GRN improves directed gene regulatory network inference from scRNA-seq data by using role-aware graph learning and cross-view contrastive regularization, achieving strong performance across multiple benchmarks.

SPSID denoises gene regulatory networks by distinguishing direct from indirect interactions, outperforming existing methods in simulations and benchmarks.

BiGAT-Fusion improves drug-disease association prediction by addressing class imbalance, directional asymmetry, and static feature fusion through a novel graph neural network model.

BOND-PEP generates controllable peptide binders using topology-conditioned bipartite alignment and retrieval-augmented methods, achieving state-of-the-art results with sequence-only inputs.

This study evaluates how large language models generate summaries of bioinformatics analyses, finding they produce readable and safe summaries but often misinterpret visualizations and lack novel insights.

The paper develops a framework to characterize and design gene panels for spatially resolved omics, showing that both computational and manually curated panels have strengths in capturing cell types and reducing redundancy.

scClassify2 is a message-passing neural network for cell state identification, outperforming existing methods in annotating adjacent cell states and applicable to multiple data types.