Protein-protein interactions are very important for many biological processes as this often leads to a particular protein complex to perform particular function. Thus, to identify different protein interactions helps to understand the function performed by that protein. The interaction between obligate and non-obligate complexes with each other is a particular problem that has drawn the attention of the research community in the past few years. In this thesis, we discuss this classification problem and show an efficient model to distinguish these two types of protein complexes correctly. We used new features such as desolvation energies for atom and amino acid type to compare with some other features which have already been used to validate and evaluate our model and test the strength of our newly selected features. We also used some well known feature selection techniques to perform classification with almost the same or higher accuracy but in time efficient manner. To achieve a better insight of this classification, we also performed some visual and post-analysis, and biochemically driven feature selection to achieve a better perspective about the reasons for interaction of these types of complexes.
展开▼
