Research On Clustering Based Gene Co-expression Pattern For Brain Regions

Transcriptional profiling of each human brain regions will be helpful for understanding the molecular mechanism of schizophrenia.In this study,each of six brain regions,including frontal lobe,temporal lobe,parietal lobe,basal ganglia,occipital lobe and hippocamel,top 250 schizophrenia candidate genes were selected,then the expression pattern of these genes were analyzed by 2D hieratical clustering method.Also,different correlation analysis was executed between each two different tissues in each sub-region.As in the tissue dimension we found that local pattern of the sample clustering can reflects the cytoarchitecture of 3 brain regions,while in other 3 the reflections are much less discriminable.As in the gene cluster dimension,we found from the GO term of each gene cluster that there were some genes have the similar co-expressing level across some brain regions,while in some other regions,the results we found is just the opposite.Our experimental results strongly proved that local pattern of schizophrenia candidate genes are not only just simply reflect sub-brain cytoarchitecture but also reflect functional coordination between each molecular components.And those co-expression pattern in all the brain regions,it may be helpful to understand the pathology mechanisms of schizophrenia related to each sub-brain.When it come to gene Co-expression pattern,traditional clustering algorithm can no longer get its way to capture the splendid biological information,therefore,we import the bi-cluster algorithms.Furthermore,We anounced a new bi-cluster algorithm which based on our former study on sub-brain co-expression pattem.This algorithm,we call it CHPBC,take CPB Algorithm as its core and PCC as its judging thread.It turns out CHPBC had a good performance on both informatively rich and biological sensablity.