Zi Wang

MetaClaw is an auditable AI agent that performs end-to-end metagenomic and multi-omics analysis, recovering key biological markers across multiple studies.

The paper identifies hidden open reading frames in plasmids using eGFP as a model, finding 6% of constructs contain hidden ORFs, mostly in the +2 frame.

LeafRank infers relative cell fitness from single-cell phylogenies in chromosomally unstable tumors, revealing fitness dynamics in ovarian cancer.

BaiZe predicts and interprets cellular responses to perturbations using a multi-view framework, showing effectiveness across diverse contexts and improved predictions with ATAC-seq data.

This paper introduces SpaWeaver, a framework for robust spatial multi-omics integration using weak anchors, showing improved performance in handling real-world challenges and enabling new biological insights.

The paper presents a verified spatial transcriptomics atlas built from public data, showing that increasing study diversity improves model transfer and reproducibility.

The paper introduces SE3CSP, an SE(3)-equivariant deep-learning framework for crystal structure prediction of nucleosides, achieving 57% accuracy and enabling the discovery of bioactive self-assembling materials.

This study identifies SMAD4 as a regulator of 3D genome architecture in lung squamous cell carcinoma, showing its role in inhibiting tumor development by constraining chromatin looping.

This study identifies 15 novel RLS-associated loci and links them to neurodevelopmental programs, while revealing over 100 comorbidities and brain structural associations through genome- and phenome-wide analyses.

The paper introduces Seq2Kd, a deep learning model predicting TF-DNA binding affinity using high-throughput Kd measurements, and provides a web portal for functional genomics and disease research.

The paper compares deep learning models with pairwise-additive models for predicting TF-DNA binding affinity using high-throughput experimental data, showing improved accuracy with non-additive models.

This paper reconstructs the ancestral Musaceae karyotype and links chromosome evolution to bract coloration diversity through genomic and transcriptomic analyses.

SpaCir-VDJ enables spatial immune repertoire profiling by integrating spatial barcodes and V(D)J sequences into circular templates, revealing clonal organization and somatic hypermutation patterns in human tissues and gastric cancer.

The paper identifies a phosphorylation switch in PAGE4 as a key mechanism in MED12-mutant fibroid pathogenesis, revealing its role in rewiring the interactome and transcriptional machinery.

QSProteome provides a platform for modeling, evaluating, and refining predicted protein complex structures, enabling community-driven exploration and refinement of over 42,000 genes.