Object Oriented Data Analysis for Untargeted Metabolomics

PROJECT SUMMARY/ABSTRACT In this Phase II SBIR application for Object Oriented Data Analysis for Untargeted Metabolomics, we propose to continue the research initiated in Phase I developing a software platform. In Phase I we showed functional Object Oriented Data Analysis (fOODA), which models raw RT x m/z x Ion Intensity data in functional form (12-14), can be applied to raw metabolomics data to automatically, and in an unbiased way, find peak differences across subject subgroups, and present the results to investigators in a clear way. The software runs fast (a few hours to process hundreds of sample files) in parallel on the AWS cloud. Requirements engineering through user discussions identified functions that will be added in Phase II (e.g., automate identification of isotopes, adducts, product ions, and ion suppression artifacts; link results to annotated databases; add analyses for dose response studies). The current software provides an initial architecture that will be further developed. Three aims are proposed: further develop statistical and annotation tools and validate them in simulations and 3 untargeted metabolomics studies (Aim 1); conduct extensive Requirements Engineering interviews with 100 metabolomics core directors, researchers, and clinicians (Aim 2); and write software to sell and install in core facilities (Aim 3). The first aim builds on the success of the Phase I project and is designed to optimize the analytic tools. The other two aims are focused on developing the commercial product. A new hire for each aim will be done to ensure we have the right skill set and expertise for these aims. The commercialization plan describes short- and long-term business development. In the short-term we will license proprietary software to metabolomics cores and offer statistical consulting services. In the long-term we will leverage our contacts with multi-omics cores to participate in BioRankings multi-omics platform where metabolomics, microbiome, and clinical data can be easily combined for translational statistical analysis. This will provide a second revenue stream for BioRankings.PROJECT NARRATIVE Untargeted metabolomics is a new technology designed to discover ‘biomarkers’ for clinical diagnosis and as targets for new treatments. This project proposes to create new statistical methods and software designed specifically for the unique structure of this data. If successful, these tools will help researchers move untargeted metabolomics research from the lab to the clinic.