Discriminating between deleterious and neutral non-frameshifting indels based on protein interaction networks and hybrid properties
- 242 Downloads
More than ten thousand coding variants are contained in each human genome; however, our knowledge of the way genetic variants underlie phenotypic differences is far from complete. Small insertions and deletions (indels) are one of the most common types of human genetic variants, and indels play a significant role in human inherited disease. To date, we still lack a comprehensive understanding of how indels cause diseases. Therefore, identification and analysis of such deleterious variants is a key challenge and has been of great interest in the current research in genome biology. Increasing numbers of computational methods have been developed for discriminating between deleterious indels and neutral indels. However, most of the existing methods are based on traditional sequential or structural features, which cannot completely explain the association between indels and the resulting induced inherited disease. In this study, we establish a novel method to predict deleterious non-frameshifting indels based on features extracted from both protein interaction networks and traditional hybrid properties. Each indel was coded by 1,246 features. Using the maximum relevance minimum redundancy method and the incremental feature selection method, we obtained an optimal feature set containing 42 features, of which 21 features were derived from protein interaction networks. Based on the optimal feature set, an 88 % accuracy and a 0.76 MCC value were achieved by a Random Forest as evaluated by the Jackknife cross-validation test. This method outperformed existing methods of predicting deleterious indels, and can be applied in practice for deleterious non-frameshifting indel predictions in genome research. The analysis of the optimal features selected in the model revealed that network interactions play more important roles and could be informative for better illustrating an indel’s function and disease associations than traditional sequential or structural features. These results could shed some light on the genetic basis of human genetic variations and human inherited diseases.
KeywordsIndel Disease Network feature Random forest Incremental feature selection
This work was supported by Grants from the National Basic Research Program of China (2011CB510102, 2011CB510101), the National Natural Science Foundation of China (61401302, 31371335, 81171342, 81201148), the Tianjin Research Program of the Application Foundation and Advanced Technology (14JCQNJC09500), the Innovation Program of the Shanghai Municipal Education Commission (12ZZ087), the National Research Foundation for the Doctoral Program of Higher Education of China (20130032120070, 20120032120073), the grant of ‘‘The First-class Discipline of Universities in Shanghai’’ and the Seed Foundation of Tianjin University (60302064, 60302069).
Conflict of interest
The authors declare that they have no conflict of interest.
- Akagi K, Stephens RM, et al (2010) MouseIndelDB: a database integrating genomic indel polymorphisms that distinguish mouse strains. Nucleic acids research 38(Database issue):D600–D606. doi 10.1093/nar/gkp1046
- Dong B, Chen J et al (2013) Two novel PRP31 premessenger ribonucleic acid processing factor 31 homolog mutations including a complex insertion-deletion identified in Chinese families with retinitis pigmentosa. Mol Vision 19:2426–2435Google Scholar