SIDEpro: A novel machine learning approach for the fast and accurate prediction of side‐chain conformations

Abstract

Accurate protein side-chain conformation prediction is crucial for protein modeling and existing methods for the task are widely used; however, faster and more accurate methods are still required. Here we present a new machine learning approach to the problem where an energy function for each rotamer in a structure is computed additively over pairs of contacting atoms. A family of 156 neural networks indexed by amino acid and contacting atom types is used to compute these rotamer energies as a function of atomic contact distances. Although direct energy targets are not available for training, the neural networks can still be optimized by converting the energies to probabilities and optimizing these probabilities using Markov Chain Monte Carlo methods. The resulting predictor SIDEpro makes predictions by initially setting the rotamer probabilities for each residue from a backbone dependent rotamer library, then iteratively updating these probabilities using the trained neural networks. After convergences of the probabilities, the side-chains are set to the highest probability rotamer. Finally, a post-processing clash reduction step is applied to the models. SIDEpro represents a significant improvement in speed and a modest, but statistically significant, improvement in accuracy when compared to the state-of-the-art for rapid side-chain prediction method SCWRL4 on the following datasets: (1) 379 protein test set of SCWRL4; (2) 94 proteins from CASP9; (3) a set of 7 large protein-only complexes; and (4) a ribosome with and without the RNA. Using the SCWRL4 test set, SIDEpro's accuracy (χ1 86.14%, χ1+2 74.15%) is slightly better than SCWRL4-FRM (χ1 85.43%, χ1+2 73.47%) and it is 7.0 times faster. On the same test set SIDEpro is clearly more accurate than SCWRL4-RRM (χ1 84.15%, χ1+2 71.24%) and 2.4 times faster. Evaluation on the additional test sets yield similar accuracy results with SIDEpro being slightly more accurate than SCWRL4-FRM and clearly more accurate than SCWRL4-RRM; however, the gap in CPU time is much more significant when the methods are applied to large protein complexes. SIDEpro is part of the SCRATCH suite of predictors and available from: http://scratch.proteomics.ics.uci.edu/.

Specific combinations of biallelic POLR3A variants cause Wiedemann-Rautenstrauch syndrome

“…In that structure, Pol III subunit RPC1, which shares 49% amino acid identity with the human homologue POLR3A (online supplementary figure 1), was used as template. After removing all side chain atoms from the complex, the backbone atom-only residues in RPC1 were formally renamed and renumbered (in PDB format) to as the corresponding ones of human POLR3A according to the pairwise sequence alignment using dedicated tools developed in-house  and parsed for side chain construction to SIDEpro 27. In the complex, residues from other subunits having direct interactions with POLR3A residues affected by the WRS-causing variants resulted to be conserved.…”

Evolutionary decision rules for predicting protein contact maps

“…This can be done in different ways, e.g., using quadratic potential GO model [3] or using tools like FT-COMAR [56,57]. It is also possible to obtain the coordinates of all protein atoms from the protein backbone using tools like SCWRL, IRECS, SCAP, SCATD or SCCOMP [19] or the recent tool SIDEpro [46]. Contact maps, as protein structure representation, are also useful to compare protein structures, using the maximum contact map overlap [13].…”

Next-Generation Sequencing Identifies Different Genetic Defects in 2 Patients with Primary Adrenal Insufficiency and Gonadotropin-Independent Precocious Puberty

“…material, see www.karger.com/doi/10.1159/000492496) (CYP11B1 alignment). Then, all side chains were added and optimized with SIDEpro [14]. The heme cofactor was added with the same binding pose as in the template structure.…”