OPUS-X: An Open-Source Toolkit for Protein Torsion Angles, Secondary Structure, Solvent Accessibility, Contact Map Predictions, and 3D Folding

Abstract

In this paper, we report an open-source toolkit for protein 3D structure modeling, named OPUS-X. It contains three modules: OPUS-TASS2, which predicts protein torsion angles, secondary structure and solvent accessibility; OPUS-Contact, which measures the distance and orientations information between different residue pairs; and OPUS-Fold2, which uses the constraints derived from the first two modules to guide folding. OPUS-TASS2 is an upgraded version of our previous method OPUSS-TASS (Bioinformatics 2020, 36 (20), 5021-5026). OPUS-TASS2 integrates protein global structure information and significantly outperforms OPUS-TASS. OPUS-Contact combines multiple raw co-evolutionary features with protein 1D features predicted by OPUS-TASS2, and delivers better results than the open-source state-of-the-art method trRosetta. OPUS-Fold2 is a complementary version of our previous method OPUS-Fold (J. Chem. Theory Comput. 2020, 16 (6), 3970-3976). OPUS-Fold2 is a gradient-based protein folding framework based on the differentiable energy terms in opposed to OPUS-Fold that is a sampling-based method used to deal with the non-differentiable terms. OPUS-Fold2 exhibits comparable performance to the Rosetta folding protocol in trRosetta when using identical inputs. OPUS-Fold2 is written in Python and TensorFlow2.4, which is user-friendly to any source-code level modification. The code and pre-trained models of OPUS-X can be downloaded from https://github.com/OPUS-MaLab/opus_x.

OPUS-Fold3: a gradient-based protein all-atom folding and docking framework on TensorFlow

“…In Table 1, we compare the backbone folding performance of OPUS-Fold3 with that of the pyRosetta folding protocol in trRosetta 8 on CAMEO60 using the identical constrains as the inputs. The results show that OPUS-Fold3 achieves comparable performance to pyRosetta either using the predicted constrains from OPUS-Contact 9 or using the real constrains derived from the corresponding PDB-file. Here, OPUS-Fold3 adopts the predicted backbone torsion angles ( Φ and Ψ ) from OPUS-TASS2 9 as its initial state.…”

“…The results show that OPUS-Fold3 achieves comparable performance to pyRosetta either using the predicted constrains from OPUS-Contact 9 or using the real constrains derived from the corresponding PDB-file. Here, OPUS-Fold3 adopts the predicted backbone torsion angles ( Φ and Ψ ) from OPUS-TASS2 9 as its initial state. When using the random values as the initial backbone torsion angles ( Φ and Ψ ), the performance is slightly decreased (OPUS-Fold3 (random) in Table 1).…”

“…In our previous study, we developed OPUS-Fold2 to respectively model the backbone 9 and side chains 10 using the distance and orientation distribution constraints proposed by trRosetta 8 . However, both folding backbone exclusively 9 and folding side chains with fixed backbone 10 have limited usages.…”

OPUS-Mut: studying the effect of protein mutation through side-chain modeling

“…For evaluation, we use three native backbone test sets: CAMEO60, collected by OPUS-Rota3 (14), contains 60 hard targets released between January 2020 and July 2020 from the CAMEO website (57); CASP14, collected by OPUS-X (9), contains 15 FM targets downloaded from the CASP website (http://predictioncenter.org); CAMEO65, collected in this study, contains 65 hard targets released between May 2021 and October 2021 from the CAMEO website.…”

“…In recent years, many protein structure prediction methods have been proposed (9–12). Among them, AlphaFold2 (11) is the best and, in some cases it delivers the predictions close to the experimental results.…”