Wenhuan Zeng

MMIP integrates microbiome and metabolome data to identify features and predict metabolite signatures from amplicon sequencing data without requiring initial metabolomic analysis.

The paper introduces MR-DNA, a deep-learning framework for predicting 5mC-methylation states in DNA sequences with 97.9% accuracy using a domain-specific loss function and a stacked model architecture.

Megan Server provides web-based interactive access to metagenomic data without requiring full data downloads.

MuLan-Methyl uses five transformer-based models to predict three types of DNA methylation sites, achieving good performance on a benchmark dataset.

DeepToA predicts the theater of activity of a microbiome using taxonomic and functional profiles with high accuracy.

The paper enhances a COVID-19 dataset by adding weather and research sentiment features, achieving 95% accuracy in survival prediction using XGBoost.