Zeyu Liang

PlasmidGPT generates engineered plasmid sequences and predicts sequence attributes using a language model trained on 153k plasmid sequences.

The paper introduces CHASING, a method for gain- or loss-of-function screening using chromosome-segment deletions, and applies it to identify metabolic engineering targets in B. subtilis.

The paper uses a language model to design cis-regulatory elements for prokaryotes without experimental data, achieving high success rates in gene expression and promoter activity.

This study develops a deep learning model to generate nonrepetitive and diverse CRISPR sgRNAs with high activity, validated for gene editing and regulation in Bacillus subtilis.