Hongyu Shen

The paper introduces RSDISC, a method using CRISPR-Cas12a for random sanitization of DNA-stored metadata, achieving 99.9% efficiency in degrading 28,258 oligonucleotides.

This paper introduces morphVQ, an automated shape analysis method that captures morphological variation using functional maps and descriptor learning, achieving comparable accuracy to manual methods and auto3DGM while being more computationally efficient.

The paper improves protein sequence model generalizability by applying data augmentations and contrastive learning, showing performance gains across multiple tasks.