Pattern Mining Method Associated With Cancer Based On Network Model

Cancer has great relationship with multiple factors such as human living environment,individual genetic factors and so on.Because of the high risk that cancer bring to human health,numbers of scientific institutions has been researching it to find the pathogenesis of it in the world wide.With the next generation sequencing technology,it’s convenient to find the important information of cancer and the relationships among them.In the 21 st century,research for cancer will tend to be integrated and multilevel,the researching point will be developing form single gene to multigene modules,marching form single-omic data towards multi-omics data,expanding from singlecell to the wholly human.The developing of system biology will inject new vitality to research for complex diseases especially cancer.System biology is not only a new research strategy,but also a new way of combating complex diseases in the future.In life science,it’s an import researching problem to mine biomarkers contributing to identify early cancer features and know the biological network regulation mechanism of cancer production and development well.Because of the extensive complexity of gene mutations in cancer genomes,identifying the key gene modules in carcinogenesis and understanding their functional mechanisms is a major challenge in the field of bioinformatics.In general,gene expression can be regulated by heterogeneous,multi-level regulatory factors such as copy numbers,methylation,transcription factors,mi RNA,etc.This paper consider multi-omics data from data integrating perspective,we introduce lnc RNA-omic data into consideration and construct biological network model,mine the cancerassociated core gene module through cluster method.We summary a systematic approach to identification core gene modules which could cause cancers.We apply this approach on LUSC and find core gene module containing 15 genes which have great relationship with cancer by analyzing their functions and pathways.We also find they can distinguish high risk group and low risk group by survival analysis.All these results show that our approach can identify core gene modules and their dysregulated genes by integrating multi-omics biological data,it benefits for cancer research.The emergence of post genomics identifies a large sets of genes.These gene sets are often associated with the same or similar disease,and are usually expressed as lists of differentially expressed genes,gene modules,protein complexes,or signaling pathways.It’s still a problem to compare gene sets and interpret their relationship into understandable potential biological mechanisms.In order to enrich the method of mining core gene modules,a density clustering method based on corrected cumulative score is proposed,this method uses the function similarity score of the gene set as the density clustering distance measure,and further excavates the core gene modules which are contained in the dysregulation gene collections by clustering.The experimental results show that the three core gene modules mined by this method are closely related to cancer,and the related functions and pathways can play an important role in cancer,which is helpful for cancer research.Finally,the platform has been built,and the training platform and clustering platform are developed to facilitate the application of the method proposed.