mzLearn: A Data-driven Metabolite Signal Detection Algorithm, bioRxiv
Zero-parameter design: No prior knowledge or QC samples are required.
Scalability: Capable of handling large-scale datasets.
First-in-class Generative Models: Learn metabolite representations linked to demographic and clinical variables, improving the accuracy of downstream predictions.
Multi-omic Data Processing: Analysis of transcriptomics, proteomics, and metabolomics data in a unified framework.
Proprietary Knowledge-Based Graph: Leverages a curated database of weighted protein-protein interactions and protein-metabolite interactions to contextualize findings.
Network Inference: Builds functional interaction networks that map multi-omic drivers of disease or drug response.