Project Summary/Abstract As a condition of the small intestine, environmental enteropathy causes to chronic inflammation and malabsorption of nutrients leading it to present itself phenotypically similar to malnutrition. Separation and identification of patients with environmental enteropathy from malnutrition can only be done via an invasive upper gastrointestinal endoscopy. In an effort to better characterize and understand environmental enteropathy, global lipidomics profiling was performed on plasma, urine, and duodenal aspirate samples of a large cohort clinical study of 415 Pakistani children. Larger sample sizes introduces technical challenges, such as the need to run samples in batches, which in turn presents computational challenges of analysis. This proposal is focused on the development of data-dependent methodologies for untargeted lipidomics analysis to identify robust lipid profiles unique to each environmental enteropathy and malnutrition. The first aim of this proposed research is to design a multi-batch analysis pipeline to efficiently and accurately aggregate data across the various batches. This pipeline will address 3 computational challenges of multi- batched data: 1) chromatographic retention time alignment, 2) missing data imputation, and 3) batch effect correction. Each of these challenges have been analyzed individually, but in data analysis each step is dependent on and influenced by the prior one. Development of an integrated, data-dependent pipeline specific to mass spectrometry data will allow for reproducible results. The pipeline will be evaluated for accuracy via testing on various sample matrices and by comparison to existing algorithms. The proposed second aim is the development of a pathway-based data-dependent tool for putative lipid identification. A bottleneck of untargeted analysis is the rapid identification of compounds. Due to the volume of data, the current approach of only identifying those features which are statistically significant creates gaps in downstream work such as pathway analysis. Introducing pathway knowledge earlier in the workflow will yield in more meaningful results. This approach uses an initial input of lipids unique to the study and builds a networks of additional connected lipids. These new lipids are stored in a database and a search is performed for them in the user’s data. Identification of lipids with this methods will lead to a more complete network profile of results. This project will identify distinctive lipidome profiles of environmental enteropathy patients and separate them from a larger malnutrition disease control cohort. This initial step will lay the foundation for future validation studies and ultimately the utilization of non-invasive diagnostics markers of environmental enteropathy, leading to improved health of these children. As large-scale studies steadily become more common and to answer the resulting computational challenges, this project will produce data-depe...