Wenpu Lai

DECIPHER improves cell-type deconvolution across molecular modalities by combining disentangled representation learning with prototype-based methods, showing robust performance in simulated and real datasets.

MIST is a deep learning framework that integrates single-cell transcriptomic and TCR data to analyze T cell function and antigen specificity, revealing heterogeneity in lung cancer CD8+ T cells and their response to immunotherapy.