ProQ3D: improved model quality assessments using deep learning

Summary: Protein quality assessment is a long‐standing problem in bioinformatics. For more than a decade we have developed state‐of‐art predictors by carefully selecting and optimising inputs to a machine learning method. The correlation has increased from 0.60 in ProQ to 0.81 in ProQ2 and 0.85 in ProQ3 mainly by adding a large set of carefully tuned descriptions of a protein. Here, we show that a substantial improvement can be obtained using exactly the same inputs as in ProQ2 or ProQ3 but replacing the support vector machine by a deep neural network. This improves the Pearson correlation to 0.90 (0.85 using ProQ2 input features). Availability and Implementation: ProQ3D is freely available both as a webserver and a stand‐alone program at http://proq3.bioinfo.se/ Contact: arne@bioinfo.se Supplementary information: Supplementary data are available at Bioinformatics online.

[1]  Pascal Benkert,et al.  QMEAN: A comprehensive scoring function for model quality assessment , 2008, Proteins.

[2]  Anna Tramontano,et al.  Methods of model accuracy estimation can help selecting the best models from decoy sets: Assessment of model accuracy estimations in CASP11 , 2016, Proteins.

[3]  Jens Meiler,et al.  ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. , 2011, Methods in enzymology.

[4]  Yann LeCun,et al.  Effiicient BackProp , 1996, Neural Networks: Tricks of the Trade.

[5]  Jie Hou,et al.  DeepQA: improving the estimation of single protein model quality with deep belief networks , 2016, BMC Bioinformatics.

[6]  Renzhi Cao,et al.  Protein single-model quality assessment by feature-based probability density functions , 2016, Scientific Reports.

[7]  Arne Elofsson,et al.  Identification of correct regions in protein models using structural, alignment, and consensus information , 2006, Protein science : a publication of the Protein Society.

[8]  Arne Elofsson,et al.  ProQ3: Improved model quality assessments using Rosetta energy terms , 2016, Scientific Reports.

[9]  J Lundström,et al.  Pcons: A neural‐network–based consensus predictor that improves fold recognition , 2001, Protein science : a publication of the Protein Society.

[10]  A. Elofsson,et al.  Can correct protein models be identified? , 2003, Protein science : a publication of the Protein Society.

[11]  Jianlin Cheng,et al.  Evaluating the absolute quality of a single protein model using structural features and support vector machines , 2009, Proteins.

[12]  Juergen Haas,et al.  The Protein Model Portal—a comprehensive resource for protein structure and model information , 2013, Database J. Biol. Databases Curation.

[13]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[14]  John Salvatier,et al.  Theano: A Python framework for fast computation of mathematical expressions , 2016, ArXiv.

[15]  Yoshua Bengio,et al.  Deep Sparse Rectifier Neural Networks , 2011, AISTATS.

[16]  Arne Elofsson,et al.  Prediction of global and local model quality in CASP7 using Pcons and ProQ , 2007, Proteins.

[17]  Karolis Uziela,et al.  ProQ2: estimation of model accuracy implemented in Rosetta , 2016, Bioinform..

[18]  Arne Elofsson,et al.  Automatic consensus‐based fold recognition using Pcons, ProQ, and Pmodeller , 2003, Proteins.

[19]  Daniel B. Roche,et al.  Assessing the quality of modelled 3D protein structures using the ModFOLD server. , 2014, Methods in molecular biology.

[20]  Matthew D. Zeiler.  ADADELTA: An Adaptive Learning Rate Method , 2012, ArXiv.

[21]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[22]  Björn Wallner,et al.  Improved model quality assessment using ProQ2 , 2012, BMC Bioinformatics.

[23]  Yoshua Bengio,et al.  Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[24]  M. Levitt,et al.  Energy functions that discriminate X-ray and near native folds from well-constructed decoys. , 1996, Journal of molecular biology.

[25]  Zheng Wang,et al.  Benchmarking Deep Networks for Predicting Residue-Specific Quality of Individual Protein Models in CASP11 , 2016, Scientific Reports.