Using N-dimensional Mutual Information to Measure Protein Multiple Residues Coevolution
Download as PDF
Hongyun Gao, Xiaoqing Yu, Shuang Chen, Dan Li, Xiaolei Zhu
Many computational tools have been developed to study amino acid co-evolution in proteins. Most of them, such as McLachlan-based substitution correlation and Mutual Information (MI), only focus on co-evolution of two amino acid residues. On the other hand, co-evolution information of multiple residues is useful to study protein structures and functions. Therefore, there is an urgent requirement for methods to identify multiple co-evolved residues. The n-dimensional Mutual Information (n DMI) method is introduced to identify co-evolution of multiple residues. This algorithm gains the co-evolution information of multiple residues from analysis of a matrix of MI values. The method was applied to G proteins. Results show that n DMI method is effective in identification of multiple co-evolved amino acids. Moreover, n DMI performs better in discovering functional important residues than methods focusing on pairwise co-evolved residues.
Mutual Information, Co-evolution, n-dimensional Mutual Information, G Proteins.