Inference Pathway Activity Based On Genes And Their Interaction Inference

The state of the cancer identification is a challenging research.It is significant for diagnosing cancer in time,designing the treatment and survival analysis.The difficulty of traditional clinical indicators such as estrogen level and progesterone is that the classification accuracy is not accurate.With the emergence of various groups of big data,such as genomic,transcriptome,proteome,metabolomics,and other modification,it is a chance for the cancer classification to develop.Based on the gene expression profile of gene transcription study,there are two types of methods: single gene analysis,the main idea is based on whole genome expression values to construct classifier,its deficiency is the consistency of cancer markers is very poor,the different data sets can't get same cancer markers.With the development of biological data such as protein network,metabolic networks,many methods mainly analyze the differences of gene expression,but they don't consider the interaction between genes.In this study,pathway activity based on genes and their interactions(PAGI)method is proposed.PAGI method is based on only one single pathway and evaluates the importance of the pathway.We incorporate the information to weight the genes and their interactions and infer pathway activity.PAGI method could make the significant genes and their interactions more remarkable,and weaken the signal of normal genes and their interactions.The strategy of weighting gene network by their importance greatly improves the reproducibility and robustness of pathway activities.We applied the PAGI method to the classification of seven types of cancers.It showed that the proposed method yielded a more accurate and robust in both within-datasets and cross-datasets experiment,compared to several existing gene-based and pathway-based classification methods.PAGI method can identify risk genes and interactions.These biology markers are more reliable in guiding therapeutic plan and helping the development of pathway-specific therapeutic strategies.