Dylan H. Ross

VaLPAS pairs a bi-directional autoencoder with correlation for omics guilt-by-association with calibrated confidence; in L. starkeyi its autoencoder linked 176 unidentified metabolites to known-function proteins at ≥50% confidence.

This paper identifies unknown molecules without reference libraries by combining computational predictions with spectrometric data, achieving exact identifications in a fentanyl case study.

The paper develops a comprehensive metabolomics database for Pseudomonas putida to support untargeted metabolomics studies and enable molecular-level analysis.

The paper proposes identification probability as an automated and transferable method to assess metabolite identification confidence in metabolomics.

The paper develops a drug exposure resource from untargeted metabolomics data to identify medications without clinical records.

The paper creates a large-scale CCS database for drugs and metabolites and develops a machine learning model to predict CCS values for different molecular forms.