A de novo protein structure prediction by iterative partition sampling, topology adjustment, and residue-level distance deviation optimization

Abstract

Motivation: With the great progress of deep learning-based inter-residue contact/distance prediction, the discrete space formed by fragment assembly cannot satisfy the distance constraint well. Thus, the optimal solution of the continuous space may not be achieved. Designing an effective closed-loop continuous dihedral angle optimization strategy that complements the discrete fragment assembly is crucial to improve the performance of the distance-assisted fragment assembly method. Results: In this article, we proposed a de novo protein structure prediction method called IPTDFold based on closed-loop iterative partition sampling, topology adjustment and residue-level distance deviation optimization. First, local dihedral angle crossover and mutation operators are designed to explore the conformational space extensively and achieve information exchange between the conformations in the population. Then, the dihedral angle rotation model of loop region with partial inter-residue distance constraints is constructed, and the rotation angle satisfying the constraints is obtained by differential evolution algorithm, so as to adjust the spatial position relationship between the secondary structures. Lastly, the residue distance deviation is evaluated according to the difference between the conformation and the predicted distance, and the dihedral angle of the residue is optimized with biased probability. The final model is generated by iterating the above three steps. IPTDFold is tested on 462 benchmark proteins, 24 FM targets of CASP13, and 20 FM targets of CASP14. Results show that IPTDFold is significantly superior to the distance-assisted fragment assembly method Rosetta_D (Rosetta with distance). In particular, the prediction accuracy of IPTDFold does not decrease as the length of the protein increases. When using the same FastRelax protocol, the prediction accuracy of IPTDFold is significantly superior to that of trRosetta without orientation constraints, and is equivalent to that of the full version of trRosetta.

Improved model quality assessment using sequence and structural information by enhanced deep neural networks

“…Various methods have been proposed to predict the threedimensional structure of proteins [1][2][3][4][5][6]. The successful application of deep learning in recent years has significantly improved the accuracy of structure prediction [7][8][9][10][11][12][13][14], especially the end-to-end deep learning methods AlphaFold2 [15], whose prediction accuracy on most of the Critical Assessment of techniques for protein Structure Prediction (CASP14) targets is close to experimental results [16].…”

Analysis of AlphaFold and molecular dynamics structure predictions of mutations in serpins

“…Furthermore, recent implementations of novel machine learning methods coupled with molecular dynamics may improve the research on the proteins' conformational ensemble [11]. As for AlphaFold, an in-depth study of the method and understanding of the code and training process might allow providing protein-targeted predictions for this family of serpin structures with higher accuracy [68].…”

GraphGPSM: a global scoring model for protein structure using graph neural networks

“…Validation of the ability to guide protein modeling. We use the above algorithm to predict models for 484 proteins of our in-house IPTDFold [10] test set. To analyze the effect of GraphGPSM in protein modeling, we design a comparative experiment that removes the GraphGPSM scoring model from the above algorithm, that is, only REF2015 [7] is used to guide protein modeling.…”

“…The non-end-to-end versions of RoseTTAFold [6] and trRosetta [9] convert deep learning predictions of inter-residue distance and orientations into bound energy potentials that guide protein modeling along with uniform field energies. IPTDFold constructs distance energy potentials to perform protein topology adjustment and local dihedral angle optimization [10]. Uniform field scoring models reflect the universal properties of proteins, but they have low accuracy, are computationally complex, and lack the unique properties of different proteins.…”

“…The non-end-to-end versions of RoseTTAFold [6] and trRosetta [9] convert deep learning predictions of inter-residue distance and orientations into bound energy potentials that guide protein modeling along with uniform field energies. IPTDFold constructs distance energy potentials to perform protein topology adjustment and local dihedral angle optimization [10].…”