Development of data driven and AI empowered systems biology to study human diseases

点击翻译按钮获取中文摘要

Project Summary Systems biology models provide an effective way to study the functional impact of biological process within complex disease system. Despite a plethora of knowledge on the differential equation-based systems biology model have gained, there are still major gaps in raising dynamic models within the context of human diseases. Essentially, the parameters involved in the non-linear dependencies are largely unknown under disease conditions and the systems biology models are always within a reductionist paradigm, which can hardly characterize the complicated disease system. The large amount of single-cell, spatial or tissue multi-omics data obtained from disease tissue has been proven to be endowed with the potential to deliver information on a cell functioning state and its underlying phenotypic switches. Hence, advanced systems biology models and computational tools are in pressing need to empower reliable characterization of biological processes and their functional roles in disease by using multi-omics data. Our preliminary data include (1) a new computational method to approximate systems biology model using transcriptomics data, and (2) computational principles to approximate dynamic system by using omics data, which form the methodology and theoretical foundations of this project. In this MIRA project, I proposed to develop a suite of novel computational methods, systems biology models and quantitative metrics to bring the following unmet capabilities: (1) A computational framework to establish dynamic models using omics data, which will enable the following analyses to study a complex disease system: (i) assessing sample-wise activity of biological processes; (ii) perturbation analysis to evaluate the impacts of biological features or model structures to the system, which could serve as new drug targets, and (iii) evaluating how the system evolve through disease progression; (2) A natural language processing-based extraction of biological functions and relations to automatically establish context specific knowledge of system structure and components from scientific literature datal; and (3) computational principles and theories of the identifiability and mathematical representation of dynamic systems in omics data. By implementing these methods into multi-omics data analysis, we plan to address the following outstanding biological questions: (i) identification of molecular features with high impact to metabolic variations in different diseases, (ii) the role of metabolism in fueling epigenetic regulation, (iii) transcriptional regulation of metabolism and other biological processes, (iv) functional annotation of genetic variations, and (v) assessment of biochemical variations. We will also develop novel knowledge representation and transfer of metabolic and other variations in pan-disease analysis to aid in better understanding of the basic disease pathology and promote the precision medicine research, including prediction and validation of new biomarkers, nutrition recommendation, and drug repurposing. Successful execution of the proposed research will provide a suite of computational capabilities to quantify and study general biological processes that could be broadly utilized by the biomedical research community.