Yiqing Yan

This paper presents a workflow using XGBoost, LightGBM, and TabNet to classify variants in tumour-only data, showing improved accuracy across ancestries and tumour types.

The paper introduces composite motifs and enzymatic techniques to improve DNA storage density and presents an end-to-end experiment achieving 84 bits/cycle logical density.

Accel-Align introduces a new sequence aligner using seed-embed-extend methodology, achieving 3-12x speed improvements over existing aligners without sacrificing accuracy.