Developing methods for high-throughput deep characterization of extracellular RNA, DNA, and vesicles in body fluids

PROJECT SUMMARY Cell turnover and active secretion release extracellular (also so-called cell-free) RNA, DNA, and vesicles into circulation. The presence of these extracellular analytes in bodily fluids has enabled broad applications in oncology, transplantation, obstetrics, and metabolic and neurological degenerative diseases. Analyzing cell-free DNA (cfDNA), cell-free RNA (cfRNA), and extracellular vesicles (EVs) provides clarity in understanding disease states for clinical management. Major technical challenges in cfDNA, cfRNA, and EV analyses are associated with their complexity, low abundance, and preanalytical variation. My lab develops new tools to address these challenges in high-throughput deep characterization of cfRNA, cfDNA, and EVs in bodily fluids. First, we will develop tools to reliably decompose and control for technical variability, RNA degradation, and complexity of cfRNA. To address technical variability, we will identify an intrinsic cfRNA reference gene set using non-negative matrix factorization. To reliably determine the cell-type-specific footprint of cfRNA, we will develop a k-mer-based classification at the individual sequencing read level, leveraging highly abundant and commonly expressed genes. To overcome the fluctuation and a high dropout rate caused by cfRNA low abundance and degradation, we will build a computational framework for anomaly detection and biomarker combination using topic modeling. Second, we will introduce a new method to induce the epigenetic contrast of cell-free nucleosomes (cf- nucleosomes) directly in body fluids through polyamine-mediated phase separation. We aim to measure and map cf-nucleosome pseudo-accessibility at sites that differ for cell states such as active and repressive chromatin marks, transcription factor binding sites (TFBSs), regulatory loci, gene bodies, and 3D genome structure. Third, we will develop a new technique to dissect the heterogeneity of EVs at the single vesicle level using proximal barcode transferring. To enhance barcode transferring efficiency, we propose using bridge amplification and DNA walking. To improve background correction and address the scarcity of proteins detected on individual EVs, we will develop a computational framework utilizing an innovative mixture model approach that accounts for noise and is robust against measurement dropout. In summary, we are developing methods to analyze low-abundance and complex circulating cfRNA, cf-nucleosomes and EVs in accessible bodily fluids. Our non-invasive approaches will enable the investigation of natural longitudinal disease progression and adaptive responses to interventions in human, non-human primates, and animal models.