Kaixuan Diao

TcrDesign generates experimentally validated, functional, full-length TCRs using deep learning, achieving state-of-the-art performance in binding prediction and sequence generation.

This study analyzes CNA signatures in single-cell and bulk samples of hepatocellular carcinoma, revealing masked patterns and potential biomarkers for prognosis and treatment.

This study uses deep learning to predict metabolic vulnerabilities in cancer cells and validates findings using TCGA data, showing improved clinical responses to targeted therapies.

The paper identifies FLAD1 as a key gene for hypoxia adaptation in tumors and develops a deep learning model to predict hypoxia-dependent metabolic genes.

This study identifies 14 CNA signatures from pan-cancer WGS and TCGA SNP array data, revealing novel patterns and prognostic implications for cancer patients.

Seq2Neo predicts neoantigen immunogenicity using a CNN model trained on genomic alterations, outperforming existing tools in independent datasets.