MicroRNA Activity Analysis And Metabolic Flux Analysis With High Throughput Data

In recent years, the rapid development of various high-throughput experiment techniques and the accumulation of abundant high-throughput omics data have brought new opportunities for the research of life science, and have also enabled us to comprehensively investigate the biological systems from different aspects and levels. However, along with the opportunities, the abundant high-throughput omics data are also bring many new challenges for the computational biologists. The challenges come from both data analysis aspect and high-throughput data modeling aspect. Focus on these challenges, we conducted the researches in this thesis.The first part of our researches was microRNA activity analysis based on gene expression data. MicroRNAs are a kind of non-coding RNA molecule found in organisms. MicroRNAs are also an important mechanism for organisms to complete transcriptional regulation, and have found to have direct relationship with many diseases and important biological processes. It’s well known that high-throughput biological omics data are of great importance to the research of microRNA regulatory mechanism. Therefore, we employed the high-throughput microarray data to investigate the activity of microRNA regulating mRNA degradation.We have proposed an approach named mirAct for microRNA activity evaluation. MirAct, with the ability of handling small data sets, takes advantage of both nonparametric method and gene expression information within a single sample and between different classes of samples. Comparing with the existing methods, mirAct enables the ability to deal with multi-class datasets. Moreover, mirAct is more robust and more competitive when facing microRNAs with low activities and data sets with intense noises.In order to serve the microRNA researcher better, mirAct is implemented as a Web Server. Comparing with the existing standalone tools, the Web Server mirAct has friendlier user interface, and provides several functions for further analysis.We employed several microRNA activity evaluating tools, including mirAct, to perform analysis on matched mRNA and microRNA expression data from different tissues and sources. Different from the intuitive expectation, we found that there are only very weak correlation between microRNA activity and expression. The results reveal the complex regulating mechanism between microRNAs and their targets.The second part of our researches was metabolic flux analysis based on high-throughput biological data. Metabolism system is the foundation of organism activities. The research on metabolism system contributes to not only the industrial production but also the clinic medicine. The development of high-throughput experiment techniques and the accumulation of high-throughput omics data offer an opportunity to investigate metabolism system in a new level. In this part, we have undertaken two sub-tasks:one for microorganisms metabolic flux analysis with Mass Spectrometry (MS) and Nuclear Magnetic Resonance (NMR) data obtained from carbon labeling experiments, the other for mammalian metabolic flux analysis with proteomics data.Carbon labeling experiment based Metabolic Flux Analysis (13C MFA), which accurately quantifies the intra-cellular metabolic flux through detecting the isotopic labeling information of metabolites, is an important technique for microorganism’s metabolism system analysis. Firstly, in order to assure the reliability of the13C MFA computation, we proposed an automatic metabolic network model verifying approach, which saves researchers from the costly manual verification. Then, based on the MS and NMR data from carbon labeling experiments, we systematically investigated the metabolic flux distribution of E. Coli. under oxidative stress. We discovered some metabolic flux shifting mechanisms that E. Coli. exploits to respond to oxidative stress. Lastly, we proposed an easy way that combines13C MFA and Flux Balance Analysis (FBA) together. Taking advantages of both the two approaches, the new method enables metabolic flux analysis in a larger metabolic network, and can effectively interpret physiologically relevant behavior.There are relatively few researches in the field of metabolic flux analysis in mammalians recently. High-throughput omics data can bring substantive additional useful information. Based on the existing method, we employed a Mix Integer Linear Programming model to combine protein differential expression information and then investigated reaction activity differences in mammalian metabolism system. The new method can avoid the thorny problem of determining highly expressed reaction with limit samples, meanwhile can demonstrate the metabolic differential effectively. With time series proteomics data, we exploited the new method to explore the metabolism differences of hepatocyte between hepatocellular carcinoma (HCC) prone transgenic mouse and normal mouse. The results demonstrate the validity of the new method and offer some new clues for HCC pathological research in metabolism level.In summary, focus on the two specific problems of microRNA activity analysis and metabolic flux analysis, this thesis conducted explorations on the field of high-throughput omics data analysis and modeling. However, the challenges arose by the high-throughput omics data haven’t been resolved at the present stage. In order to figure out a better way to respond to the challenges, the computational biology researchers and experimental biologists should cooperate more closely. We believe that this field will still be an excellent platform for the communication between information science and life science for a long time in the future.