Object Oriented Data Analysis for Untargeted Metabolomics

PROJECT SUMMARY/ABSTRACTIn this Phase I SBIR application, we propose to develop and optimize functional Object Oriented Data Analysis(fOODA) methods and create a cloud-based SaaS platform to analyze untargeted metabolomics data.Untargeted metabolomics quantifies the amount of known and unknown metabolites in samples with thepurpose of finding known and unknown metabolites that correlate with subgroups (e.g., diseased or healthytissue). The goal of this research is to identify biomarkers for diagnosis and treatment targets. However, thebiostatistical tools being used have not been optimized for this data, and so analyses lag behind the technicaldevelopment.Functional OODA is an emerging area of statistics for analyzing functional data, such as untargetedmetabolomic retention time x m/z intensity data metabolic data. The goal of this project is to optimize fOODAfor this data to provide more powerful statistical methods for investigators. This should help them identifymetabolite biomarkers more efficiently and avoid incorrectly calling unimportant metabolites as clinicallyrelevant. Incorrect or inferior analyses results in loss of money and time running downstream experiments tovalidate these false positive calls.The innovation of this proposal is three-fold: (1) development of statistical models and methods specificallydesigned and optimized for raw (not pre-processed) untargeted metabolomics intensity data, (2) developmentof methods that allow inclusion of metadata and other omics data to measure how these factors impactmetabolomic profiles, and (3) implementation of the methods in a cloud-based SaaS solution. The impact onmetabolomics will be new tools for rigorous experimental design, hypothesis testing, and biostatistical analysisfor discovery of disease biomarkers, and to move preclinical discoveries from the lab to the clinic faster.BioRankings has filed patent protection for this technology.