Bioinformatics And Systems Biology Analysis Of Stem Cell Expression Profiles

Embryonic stem cells (ESCs) are pluripotent and have great potential in medicine. The induced pluripotent cells (iPSCs) are derived from somatic cells to mimic ESCs through the ectopic expression of several transcription factors. Human iPSCs can be generated from patient's own cell to avoid the ethic and immunological issues which raised concerns in the use of ESCs. Stem cell differentiation and the maintenance of self-renewal are intrinsically complex processes requiring the coordinated dynamic expression of hundreds of genes and proteins in precise response to external signal cues. At present, the research of dissecting this complexity is a very active emerging domain of stem cell system biology.In this study, we adoptd the concept of "state space" from the theory of system dynamics and treat the gene expression profiles as state vectors. At system level, we focused on the system behavior of gene regulatory network embedded in the gene expression profiles. We initially established the system platform for the analysis of gene expression profiles of stem cells with the self-organizing mapping neural network as well as the statistical complexity measure. The self-organizing mapping neural network is used to discover and exhibit the data distribution of the gene expression profile while the statistical complexity measure is used to depict the nonlinear dynamic characteristics of the gene regulatory network. The main contents of this thesis are as follows:(1) We used self-organizing maps to discover and exhibit the expression patterns in the gene expression profiles of stem cell. After comparative analyses, we found that there is a ground state of stem cell pluripotency which attracts various iPSCs derived from different cell types. Our observations can be regarded as an evidence of the hypothesis of the pluripotent attractor.(2) We initially revealed that the dynamics of PluriNet are closely associated with the global expression pattern of hiPSCs. Through an integrated investigation of molecular network and gene expression profiles together, our results suggested the expression pattern or the core attractor state in stem cell should be the result of the dynamic interaction of a pluripotent network.(3) We applied the technology of network description of lossless compression to the pluripotent network containing 20 core transcription factors. The results indicated that the network description of lossless compression reduce network complexity by explicitly representing re-occurring network motifs and lead to new insights to the role of these reprogramming transcription factors.(4) We put forward the statistical complexity measure based on the self-organizing maps and programmed it in MATLAB. After the quantitative analyses of the expression patterns of stem cells, our results showed that the statistical complexity measure can depict the change of cell states during reprogramming, which implies the potential value of the statistical complexity measure in gene expression analysis.(5) In a unique perspective, we discussed the origin of cancer stem cell with qualitative comparison between self-organizing maps combined with the quantitative analysis of the statistical complexity measure. Our results supported the hypothesis that the LSCs origin from HSCs. In addition, it is entirely possible that the non-tumorigenic tumor cells can be transformed to cancer stem cells in tumor.