Limsoon Wong

This paper evaluates LLM capabilities for analyzing genome-scale metabolic models, identifying strengths and limitations across domain knowledge, flux prediction, pathway construction, and optimization tasks.

Skiver estimates error rates and quality scores in metagenomic datasets using (k, v)-mer sketches without reference genomes or alignment.

The paper benchmarks 9 somatic variant calling algorithms on cfDNA samples from cancer patients, identifying optimal methods based on ctDNA levels and sequencing depth.

This paper introduces Bidirectional Beam Search for efficient trace reconstruction in DNA storage systems, achieving high accuracy and speed on real datasets.

GCfix corrects GC bias in cell-free DNA data across diverse coverages, outperforming existing methods with two new performance metrics.

This paper benchmarks eleven computational tools for DNA-binding protein identification, highlights limitations, and recommends improvements, providing a consensus method and datasets.

Fragle quantifies ctDNA from fragment length profiles using deep learning, showing high accuracy and improved detection limits across multiple datasets and sequencing types.

HydRA predicts RNA-binding capacity using protein interaction and sequence data, identifying novel RNA-binding domains with high specificity and sensitivity.

The paper demonstrates that analyzing proteomic data directly at the MS level using MZ-Bin improves class separation and highlights relevant protein splice forms over canonical sequences.

This paper introduces rank-based network approaches to improve analytical reliability in clinical proteomics, showing they outperform traditional methods in precision, stability, and cross-validation.