Jake Y. Chen

This study uses human iPSC-derived skin organoids to model fibrosis, identifies RUNX2 as a key driver, and finds a small molecule that suppresses fibrotic programs.

This paper identifies two separable KRAS-combination partners in pancreatic cancer: an induced ERBB2/3 up-regulation and a standing FAK dependency, both computationally nominated from public data.

The paper generates a CRISPRi perturbation cell atlas in human induced pluripotent stem cells, identifies novel regulators of pluripotency and metabolism, and proposes a machine learning framework for predicting perturbation recipes.

MondrianMap provides multi-resolution visualization of gene set enrichment results, revealing directional oppositions and temporal dynamics across biological processes in cancer, aging, and exercise data.

PAG-Agent performs pathway-level analysis and interpretation of transcriptomic data, identifying biologically relevant pathways and generating hypotheses across datasets and conditions.

This study benchmarks compact transformer models for de novo antibody design, finding no significant architectural differences in performance at small scales.

The paper describes the Common Fund Data Ecosystem (CFDE), which integrates NIH Common Fund data resources to enable discovery, analysis, and collaboration across heterogeneous datasets.

GETgene-AI prioritizes actionable cancer drug targets by integrating network-based methods, machine learning, and literature analysis, achieving higher precision and efficiency than existing tools.

LlamaAffinity predicts antibody-antigen binding affinity using a Llama 3 backbone and antibody sequences from the OAS database, achieving high accuracy and efficiency.

GOALS introduces a novel framework for analyzing Gene Ontologies by discretizing them into optimal layers to enhance functional insights and visualization in biological processes.

This paper develops a neuro-symbolic AI model, NeSyDPP4-QSAR, to predict DPP-4 inhibitor activity, achieving high accuracy in IC50 prediction for drug discovery.

PETS simulates drug-gene interactions using multi-scale network propagation to identify and rank drug candidates for reversing disease-specific gene expression signatures.

This paper develops a neuro-symbolic AI model using LTN to predict hERG inhibition, outperforming existing models on benchmark datasets.

This paper introduces the Mondrian Map, a visualization tool for differential pathway analysis, and demonstrates its application in glioblastoma multiforme to reveal pathway patterns across disease stages.

The paper defines AI-readiness criteria for biomedical data, encompassing seven dimensions, and evaluates them across flagship datasets to ensure ethical and explainable AI practices.

The paper introduces todenE, a clustering method combining topology and density-based approaches to group PAGs into Super-PAGs, improving functional representation in genomics.

This paper introduces MLPA, a digital twin framework for personalized cancer simulation and treatment optimization, showing high predictive accuracy in tumor growth and drug response.

BioRSP quantifies enrichment patterns of cell subsets in single-cell embeddings using polar scanning, providing interpretable summaries and stability diagnostics.

The paper introduces the CM4AI project, which aims to create AI-ready datasets of human cell architecture using multimodal data from cell lines.

This paper introduces GTKM, a method for visualizing gene expression data in glioblastoma, showing it identifies gene patterns linked to clinical outcomes and subtypes.

The paper introduces GRIPP, a framework to prioritize genes from multiple data modalities in GBM, showing iterative expansion and ranking can improve gene lists but may reduce modality-specific significance.

The paper introduces the Polar Gini Curve technique to detect single-cell biomarkers by combining gene expression and 2D visual information, identifying potential novel cardiac muscle cell subtypes in neonatal mouse heart data.