Identification Of Pattern Genes Based On Serial Transcriptomes And Its Associations With Tissue Functions

Spatiotemporal variation of gene expression can happen extensively among tissues, developmental stages and physiological conditions. The variation is associated with gene function and pathology. Benefiting from current applications of high-throughput technologies, e.g. microarray and next generation sequencing, simultaneously monitoring gene differential expressions in large scale becomes easier. From these high-throughput data, genes with particular expression patterns can be identified.Pattern genes are defined as genes that exhibit modularized expression behavior under serial physiological conditions. Four types of pattern genes are presently attracting significant attention. They are specific genes, selective genes, housekeeping genes, and repressed genes. Those genes will provide a shortcut toward a global and dynamic understanding of gene functions and their roles in particular biological events like development and pathogenesis. Upon these high-throughput data, various methods were adopted previously on describing and detecting such pattern genes. However, most of them were short of reliable and quantitative evaluation of pattern genes. Here, we designed four statistical parameters, i.e. Specificity Measure (SPM), Contribution Measure (CTM), Dispersion Measure (DPM) and Repression Measure (RPM), to aid identification and quantitative evaluation of those four types of pattern genes from serial transcriptomes.Accordingly, based on the identification method of pattern genes, a comprehensive Pattern Gene Analysis Platform was constructed. The platform includes an on-line server PaGeFinder (http://bioinf.xmu.edu.cn/PaGeFinder/) and a database PaGenBase (http://bioinf.xmu.edu.cn/PaGenBase/). The PaGeFinder provides friendly user interface for interactive identification of pattern genes from custom-submitted dataset of serial transcriptomes. PaGenBase deposits 906,599 pattern genes identified from the literature and data mining of 143 high-throughput datasets, which involve 11 model organisms,119,538 genes,1,062 samples and 1,145,277 gene expression profiles. These two on-line platforms will serve as valuable resources for global and dynamic understanding of gene behavior under various spatiotemporal environments. They will help to discover potential biomarkers, novel drug targets and molecular controls.Human body is the most complex multicellular organism. Tissue research plays a key role in understanding the structure of human body and curing diseases. We integrated tissue-related pattern genes in PaGenBase. We found that functional encrichments of tissue-specific genes and housekeeping genes are different. We studied not only the tissue specificity, but also the tissue connection. It was the first time that tissue connection was studied in the aspect of function, structure and development in a large scale. To further study, we combined the connections between diseases/chemicals and tissues, and found some tissue-specific/selective diseases/chemicals. Our research will help us understand the nature of human tissues, and provide clues on mechanism and therapy of diseases.