Prajit Rajkumar

Describes use of retrieval-augmented language models to classify chemical source information at scale within metabolomics workflows.

A Nature Protocols guide describing a four-part reverse metabolomics workflow for mining tandem mass spectrometry (MS/MS) data from public repositories and linking spectra to phenotype-relevant metadata.

Presents the GNPS Drug Library: an MS/MS-based resource integrating drug and metabolite spectra with structured pharmacologic metadata to enable empirical detection of drug exposure from untargeted metabolomics data.

Examines the impact of missing data and imputation methods (kNN, Random Forest) on untargeted metabolomics analyses and the effects of compositional data approaches on interpretation.