![]() ![]() Since the contact-map can dictate the global topology of protein structure, accurate prediction of contact-maps from the primary sequence can have important impact on the computational folding of protein structures, in particular to the proteins that lack homologous templates in the Protein Data Bank (PDB) ( Zhang, 2008). ![]() For a protein sequence with L residues, the physical contacts of all its residues pairs can be represented as a sparse L × L symmetric matrix called ‘contact-map’ the entry of the contact-map equals to 1 if the corresponding residues pair is in contact or 0 otherwise. The functions of proteins are essentially determined by the unique 3D structures, which are, from the view point of physics, formed and stabilized by direct interactions of atoms, termed ‘contacts’. Proteins are the focus of many areas of life science studies as they are responsible for most of the biological functions in living organisms. The standalone package and online server of ResPRE are made freely available, which should bring important impact on protein structure and function modeling studies in particular for the distant- and non-homology protein targets. It was also found that appropriate collection of MSAs can further improve the accuracy of final contact-map predictions. Meanwhile, the residual network with parallel shortcut layer connections increases the learning ability of deep neural network training. Detailed data analyses show that the major advantage of ResPRE lies at the utilization of precision matrix that helps rule out transitional noises of contact-maps compared with the previously used covariance matrix. The approach was tested on a set of 158 non-homologous proteins collected from the CASP experiments and achieved an average accuracy of 50.6% in the top- L long-range contact prediction with L being the sequence length, which is 11.7% higher than the best of other state-of-the-art approaches ranging from coevolution coupling analysis to deep neural network training. We developed a new method, ResPRE, to predict residue-level protein contacts using inverse covariance matrix (or precision matrix) of multiple sequence alignments (MSAs) through deep residual convolutional neural network training. ![]()
0 Comments
Leave a Reply. |