摘要

Using the topological structure of the Gene Ontology hierarchical scheme and the local and global semantics impacts of the protein function terms, this paper proposes a new measure for protein function similarity and another new measure for protein similarity. Based on these measures and the assumption that unknown proteins should share the most frequent functions of their neighbor known proteins, a new protein function prediction algorithm is proposed. We conducted experiments to compare our algorithm with other similar algorithms on real data sets and the results analysis shows that our new algorithm has some advantages in precision, recall and F-value evaluation.

  • 出版日期2012

全文