Cross-category Term Similarity Algorithm Research In Gene Ontology

With the development of biological science and technology,the total number of biological data and data complexity is high-speed growing.For the same kind of biological data,different biologists may use different ways of describing,and it can lead to other people's misunderstanding of biological data.To solve the confusing phenomenon of biological definition,make all kinds gene product functional description in databases consistent,keep high consistency in different biological database query,genome researchers organized a professional association known as the gene ontology(GO)union.Gene ontology is a very popular language,which describes the classification and properties of biological entity.In order to automatically find new relationships between the gene ontology,gene ontology similarity calculation method has been widely studied,and it is still very active in the field of semantic comparison and search.There is a certain amount of research in the field of cross-category term similarity algorithm at home and abroad,and some algorithms are developed to calculate the similarity.So far,the algorithms can be roughly divided into two categories: the algorithm based on VSM(vector space model),but the similarity is undirected;the algorithm based on ASR(association rules),but shallow annotation problem is ignored by the algorithm.In order to compute cross-category term similarity of the gene ontology more accurately and provide biomedical researchers with more reliable algorithm and data of biological information,this paper mainly studied from two aspects: First,we studied the gene ontology data,the construction of gene ontology,and statistical knowledge to compute the similarity.Our algorithm resolved the problem of direction and shallow annotation.We compare our algorithm with the algorithm based on VSM,the algorithm based on ASR and Cro GO after we designed the new cross-category term similarity algorithm of the gene ontology,and we prove the advantage of our algorithm.Even when the gene functional network are incomplete,our algorithm still has higher accuracy rate.Second,we used the algorithm to construct the association network of gene ontology term.What's more,we proved that our algorithm is more practical.Through the analysis of gene ontology term network,we proved that the cross-category term similarity algorithm studied by us is practical.The network is able to provide other researchers with a platform to use the GO data.