Structural coordinates: A novel approach to predict protein backbone conformation

Abstract

Motivation Local protein structure is usually described via classifying each peptide to a unique class from a set of pre-defined structures. These classifications may differ in the number of structural classes, the length of peptides, or class attribution criteria. Most methods that predict the local structure of a protein from its sequence first rely on some classification and only then proceed to the 3D conformation assessment. However, most classification methods rely on homologous proteins’ existence, unavoidably lose information by attributing a peptide to a single class or suffer from a suboptimal choice of the representative classes. Results To alleviate the above challenges, we propose a method that constructs a peptide’s structural representation from the sequence, reflecting its similarity to several basic representative structures. For 5-mer peptides and 16 representative structures, we achieved the Q16 classification accuracy of 67.9%, which is higher than what is currently reported in the literature. Our prediction method does not utilize information about protein homologues but relies only on the amino acids’ physicochemical properties and the resolved structures’ statistics. We also show that the 3D coordinates of a peptide can be uniquely recovered from its structural coordinates, and show the required conditions under various geometric constraints.

Effective Local and Secondary Protein Structure Prediction by Combining a Neural Network-Based Approach with Extensive Feature Design and Selection without Reliance on Evolutionary Information

“…In our recent work [16], we proposed using the same protein blocks as cluster centers, but we assigned cluster labels based on the root mean square deviation (RMSD) [17] distance instead of the RMSDA. The RMSD for protein blocks can be written as follows:…”

“…., 3M denote the Cartesian coordinates of the backbone atoms N, Cα, and C, respectively, of the M-residue protein block PB k (M = 5 for the PBs as in (1)), whereas the minimum is taken over by all of the spatial colocalization of the two protein blocks PB 1 and PB 2 . By definition, an RMSD is close to zero only when two structures are identical in three dimensions, and using the RMSD may therefore be preferable to using the RMSDA, which does not always satisfy this criterion [16]. However, RMSD calculations require assessing all possible alignments between two query fragments, and thus, they have higher computational costs than RMSDA calculations.…”

“…In the context of 3D coordinate reconstruction, cluster centers are treated as basic structures, and distances from the query fragment to these basic structures are referred to as "structural coordinates". We previously demonstrated that, given a large enough set of basic structures, one can reconstruct the coordinates of the backbone atoms of a protein fragment [16]. The number of basic structures required for such a reconstruction depends on the number of degrees of freedom (i.e., bond angles, bond lengths, and dihedral angles).…”

“…For instance, for a five-residue fragment with all bond lengths and bond angles considered to be fixed, one would need at least twelve basic structures to reconstruct the conformation of the backbone. We previously developed and implemented an algorithm [16] that could unambiguously reconstruct a protein's backbone coordinates from its "structural coordinates" representation.…”

“…In our previous work [16], we did not utilize evolutionary information. Instead, our feature encoding relied solely on amino acids' physicochemical properties and resolved structures' statistics.…”