Protein Depth Calculation and the Use for Improving Accuracy of Protein Fold Recognition

Abstract

Protein structure and function are largely specified by the distribution of different atoms and residues relative to the core and surface of the molecule. Relative depths of atoms therefore are key attributions that have been widely used in protein structure modeling and function annotation. However, accurate calculation of depth is time consuming. Here, we developed an algorithm which uses Euclidean distance transform (EDT) to convert the target protein structure into a 3D gray-scale image, where depths of atoms in the protein can be conveniently and precisely derived from the minimum distance of the pixels to the surface of the protein. We tested the proposed EDT-based method on a set of 261 non-redundant protein structures, which shows that the method is 2.6 times faster than the widely used method proposed by Chakravarty and Varadarajan. Depth values by EDT method are highly accurate with a Pearson's correlation coefficient ≈1 compared to the calculations from exhaustive search. To explore the usefulness of the method in protein structure prediction, we add the calculated residue depth to the scoring function of the state of the art, profile-profile alignment based fold-recognition program, which shows an additional 3% improvement in the TM-score of the alignments. The data demonstrate that the EDT-based depth calculation program can be used as an efficient tool to assist protein structure analysis and structure-based function annotation.

Analyzing Residue Surface Proximity to Interpret Molecular Dynamics

“…In this work, we have integrated existing algorithms [TA96] and ideas [KB15, XLZ13] to a novel measurement, Residue Surface Proximity, which reflects the closeness of the atoms or residues of a molecule to its surface or core. The novel aspect is that this is done with respect to virtual layers of the Solvent Accessible Surface definition and applied to MD simulation trajectories, providing a domain expert with a quick overview over the whole trajectory.…”

“…Such advances can also be found in the chemical domain, e.g., Saunders et al [SB02] infer information from the relative surface participation of individual residues. Further, Xu et al [XLZ13] compute the distance of an atom to the SAS. This magnitude reflects how deep an atom is buried inside a molecule and it is shown to be an interesting, additional feature, besides RMSD or RG.…”

“…In this way, we obtain a scalar value ℒ o ∊ [0, N − 1] ∊ IR for each residue and each time‐step. We see this approach as a trade off between Xu et al [XLZ13] and Karampudi and Bahadur [KB15]. While the algorithm by Xu et al yields a precise distance value of each atom to the surface, it neglects the SAS property that we maintain through the layer approach.…”

XSuLT: a web server for structural annotation and representation of sequence-structure alignments

“…Whereas SA refers to the surface of a residue exposed to the bulk solvent, depth relates to the distance of an atom or residue to that surface, allowing the quantitative discrimination between residues that lie just under the surface of the protein from those deep in its core. XSuLT can use the program EDTSurf ( 27 ) to calculate the average depth of each residue, and applies the value as a gradient of colour grey on the background of each residue, with darker shades indicating greater depth.…”

Prediction of structural features and application to outer membrane protein identification

“…The concept of RD supplements the information provided by RSA. The RD values of proteins were calculated by EDTSurf 42 program. The values output by EDTSurf lie in [2.8, 9.8], where a higher value corresponds to a deeper region where a residue locate.…”