Improved model quality assessment using ProQ2

By Arjun Ray, Erik Lindahl, Björn Wallner. 2012

193 citations recorded by the source corpus.

Employing methods to assess the quality of modeled protein structures is now standard practice in bioinformatics. In a broad sense, the techniques can be divided into methods relying on consensus prediction on the one hand, and single-model methods on the other. Consensus methods frequently perform very well when there is a clear consensus, but this is not always the case. In particular, they frequently fail in selecting the best possible model in the hard cases (lacking consensus) or in the easy cases where models are very similar. In contrast, single-model methods do not suffer from these drawbacks and could potentially be applied on any protein of interest to assess quality or as a scoring function for sampling-based refinement. Here, we present a new single-model method, ProQ2, based on ideas from its predecessor, ProQ. ProQ2 is a model quality assessment algorithm that uses support vector machines to predict local as well as global quality of protein models. Improved performance is obtained by combining previously used features with updated structural and predicted features. The most important contribution can be attributed to the use of profile weighting of the residue specific features and the use features averaged over the whole model even though the prediction is still local. ProQ2 is significantly better than its predecessors at detecting high quality models, improving the sum of Z-scores for the selected first-ranked models by 20% and 32% compared to the second-best single-model method in CASP8 and CASP9, respectively. The absolute quality assessment of the models at both local and global level is also improved. The Pearson’s correlation between the correct and local predicted score is improved from 0.59 to 0.70 on CASP8 and from 0.62 to 0.68 on CASP9; for global score to the correct GDT_TS from 0.75 to 0.80 and from 0.77 to 0.80 again compared to the second-best single methods in CASP8 and CASP9, respectively. ProQ2 is available at http://proq2.wallnerlab.org.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

DockQ: A Quality Measure for Protein-Protein Docking Models (2016, methodology): “These methods have been successful in CASP for predicting the quality of protein structure models [14,15] and there is no reason to believe that they will not be as successful in predicting the quality of docking models.”
Improved protein structure refinement guided by deep learning based accuracy estimation (2020, background): “Then a pool of structures similar to this structure (S-score (31) > 0.”
Local Error Estimates Dramatically Improve the Utility of Homology Models for Solving Crystal Structures by Molecular Replacement (2015, methodology): “In principle, it would be possible to calculate X-ray data for structures determined by other methods, and simulate an X-ray structure that could participate in LLG scoring.”
ProQ3D: improved model quality assessments using deep learning (2016, methodology): “In ProQ2 profile weights were added to improve the predictions (Ray et al., 2012) and in ProQ3 (Uziela et al., 2016) we added energy terms calculated from Rosetta (Leaver-Fay et al., 2011).”
Comparative Protein Structure Modeling Using MODELLER (2016, methodology): “Some of the programs that can be used to detect suitable templates are FFAS (Jaroszewski et al., 2005), SP3 (Zhou and Zhou, 2005), SALIGN (Marti-Renom et al., 2004), HHBlits (Remmert et al., 2012), HHsearch (Soding, 2005), and PPSCAN, part of MODELLER (Sali and Blundell, 1993).”
Deep convolutional networks for quality assessment of protein folds (2018, methodology): “The method has been reported to outperform ProQ2 [10], which was the top-performing method in the CASP11 competition [3].”
Improved de novo structure prediction in CASP11 by incorporating coevolution information into Rosetta (2016, methodology): “They also thank the CASP11 organizers, the structural biologists who generously provided targets, and Jinbo Xu (RaptorX), Johannes S € oding (HHSearch), and Yaoqi Re-ranking the five submitted models using a single-model MQA method, ProQ2, improves model 1 accuracy for FM domains.”
VoroMQA: Assessment of protein structure quality using interatomic contact areas (2017, methodology): “(5) we employed the probabilities of the contact categories that were estimated from the set of lower quality structures.”
GraphQA: protein model quality assessment using graph convolutional networks (2020, unknown): “The field of bioinformatics, on the other hand, has witnessed noticeable progress in QA for more than a decade: from earlier works using artificial neural networks (Wallner & Elofsson, 2006) or support vector machines (Ray et al., 2012; Uziela et al., 2016) to more recent deep learning methods based”
CRITICAL ASSESSMENT OF TECHNIQUES FOR PROTEIN STRUCTURE PREDICTION (2014, methodology): “We used the FAMS[1] as the homology modeling program, PSI-BLAST, and raptorX[2].”
Protein model quality assessment using 3D oriented convolutional neural networks (2018, background): “It is also very close to the performance of composite meta-methods that also use sequence alignment and evolutionary information as additional features (18).”
Methods for estimation of model accuracy in CASP12 (2017, methodology): “For the local/per-residue error estimates, each model was considered individually using 2 new pure-single model methods, the Contact Distance Agreement and the Secondary Structure Agreement scores,(16) as well as the best pure single method in earlier CASPs, ProQ2.(21,40) Additionally, 3 alternative”
Smooth orientation-dependent scoring function for coarse-grained protein quality assessment (2018, methodology): “In addition, single-model QA methods are proved to achieve better performance compared to consensus-model QA methods on unbalanced protein model pools and in cases where protein models within assessed pool are very similar (21).”
Protein Modelling: What Happened to the “Protein Structure Gap”? (2013, unknown): “, 1993) or apply a combination of knowledge-based statistical measures derived from high-resolution crystals structures to provide estimates of model accuracy (Benkert et al., 2011; Ray et al., 2012; Wiederstein and Sippl, 2007).”