Analysis of several key factors influencing deep learning-based inter-residue contact prediction

Abstract Motivation Deep learning has become the dominant technology for protein contact prediction. However, the factors that affect the performance of deep learning in contact prediction have not been systematically investigated. Results We analyzed the results of our three deep learning-based contact prediction methods (MULTICOM-CLUSTER, MULTICOM-CONSTRUCT and MULTICOM-NOVEL) in the CASP13 experiment and identified several key factors [i.e. deep learning technique, multiple sequence alignment (MSA), distance distribution prediction and domain-based contact integration] that influenced the contact prediction accuracy. We compared our convolutional neural network (CNN)-based contact prediction methods with three coevolution-based methods on 75 CASP13 targets consisting of 108 domains. We demonstrated that the CNN-based multi-distance approach was able to leverage global coevolutionary coupling patterns comprised of multiple correlated contacts for more accurate contact prediction than the local coevolution-based methods, leading to a substantial increase of precision by 19.2 percentage points. We also tested different alignment methods and domain-based contact prediction with the deep learning contact predictors. The comparison of the three methods showed deeper sequence alignments and the integration of domain-based contact prediction with the full-length contact prediction improved the performance of contact prediction. Moreover, we demonstrated that the domain-based contact prediction based on a novel ab initio approach of parsing domains from MSAs alone without using known protein structures was a simple, fast approach to improve contact prediction. Finally, we showed that predicting the distribution of inter-residue distances in multiple distance intervals could capture more structural information and improve binary contact prediction. Availability and implementation https://github.com/multicom-toolbox/DNCON2/. Supplementary information Supplementary data are available at Bioinformatics online.

Download Full-text

Enhancing coevolution-based contact prediction by imposing structural self-consistency of the contacts

10.1101/279190 ◽

2018 ◽

Author(s):

Maher M. Kassem ◽

Lars B. Christoffersen ◽

Andrea Cavalli ◽

Kresten Lindorff-Larsen

Keyword(s):

False Positive ◽

Structure Prediction ◽

Structural Information ◽

Protein Structures ◽

Sequence Alignments ◽

Homologous Proteins ◽

Multiple Sequence ◽

Contact Prediction ◽

Interaction Sites ◽

Self Consistency

AbstractBased on the development of new algorithms and growth of sequence databases, it has recently become possible to build robust and informative higher-order statistical sequence models based on large sets of aligned protein sequences. By disentangling direct and indirect effects, such models have proven useful to assess phenotypic landscapes, determine protein-protein interaction sites, and in de novo structure prediction. In the context of structure prediction, the sequence models are used to find pairs of residues that co-vary during evolution, and hence are likely to be in spatial proximity in the functional native protein. The accuracy of these algorithms, however, drop dramatically when the number of sequences in the alignment is small, and thus the highest ranking pairs may include a substantial number of false positive predictions. We have developed a method that we termed CE-YAPP (CoEvolution-YAPP), that is based on YAPP (Yet Another Peak Processor), which has been shown to solve a similar problem in NMR spectroscopy. By simultaneously performing structure prediction and contact assignment, CE-YAPP uses structural self-consistency as a filter to remove false positive contacts. At the same time CE-YAPP solves another problem, namely how many contacts to choose from the ordered list of covarying amino acid pairs. Our results show that CE-YAPP consistently and substantially improves contact prediction from multiple sequence alignments, in particular for proteins that are difficult targets. We further show that CE-YAPP can be integrated with many different contact prediction methods, and thus will benefit also from improvements in algorithms for sequence analyses. Finally, we show that the structures determined from CE-YAPP are also in better agreement with those determined using traditional methods in structural biology.Author summaryHomologous proteins generally have similar functions and three-dimensional structures. This in turn means that it is possible to extract structural information from a detailed analysis of a multiple sequence alignment of a protein sequence. In particular, it has been shown that global statistical analyses of such sequence alignments allows one to find pairs of residues that have covaried during evolution, and that such pairs are likely to be in close contact in the folded protein structure. Although these insights have led to important developments in our ability to predict protein structures, these methods generally result in many false positive contacts predicted when the number of homologous sequences is not large. To deal with this issue, we have developed CE-YAPP, a method that can take a noisy set of predicted contacts as input and robustly detect many incorrectly predicted contacts within these. More specifically, our method performs simultaneous structure prediction and contact assignment so as to use structural self-consistency as a filter for erroneous predictions. In this way, CE-YAPP improves contact and structure predictions, and thus advances our ability to extract structural information from analyses of the evolutionary record of a protein.

Download Full-text

AttentiveDist: Protein Inter-Residue Distance Prediction Using Deep Learning with Attention on Quadruple Multiple Sequence Alignments

10.1101/2020.11.24.396770 ◽

2020 ◽

Author(s):

Aashish Jain ◽

Genki Terashi ◽

Yuki Kagaya ◽

Sai Raghavendra Maddhuri Venkata Subramaniya ◽

Charles Christoffer ◽

...

Keyword(s):

Deep Learning ◽

Structure Prediction ◽

Prediction Models ◽

3D Structure ◽

Evolutionary Information ◽

Sequence Alignments ◽

Multiple Sequence ◽

Contact Prediction ◽

Multiple Sequence Alignments ◽

Distance Prediction

ABSTRACTProtein 3D structure prediction has advanced significantly in recent years due to improving contact prediction accuracy. This improvement has been largely due to deep learning approaches that predict inter-residue contacts and, more recently, distances using multiple sequence alignments (MSAs). In this work we present AttentiveDist, a novel approach that uses different MSAs generated with different E-values in a single model to increase the co-evolutionary information provided to the model. To determine the importance of each MSA’s feature at the inter-residue level, we added an attention layer to the deep neural network. The model is trained in a multi-task fashion to also predict backbone and orientation angles further improving the inter-residue distance prediction. We show that AttentiveDist outperforms the top methods for contact prediction in the CASP13 structure prediction competition. To aid in structure modeling we also developed two new deep learning-based sidechain center distance and peptide-bond nitrogen-oxygen distance prediction models. Together these led to a 12% increase in TM-score from the best server method in CASP13 for structure prediction.

Download Full-text

A novel sequence alignment algorithm based on deep learning of the protein folding code

Bioinformatics ◽

10.1093/bioinformatics/btaa810 ◽

2020 ◽

Cited By ~ 1

Author(s):

Mu Gao ◽

Jeffrey Skolnick

Keyword(s):

Protein Folding ◽

Deep Learning ◽

Sequence Alignment ◽

Protein Sequence ◽

Protein Structures ◽

Supplementary Information ◽

Alignment Algorithm ◽

Sequence Alignments ◽

Alignment Algorithms ◽

Structural Alignments

Abstract Motivation From evolutionary interference, function annotation to structural prediction, protein sequence comparison has provided crucial biological insights. While many sequence alignment algorithms have been developed, existing approaches often cannot detect hidden structural relationships in the ‘twilight zone’ of low sequence identity. To address this critical problem, we introduce a computational algorithm that performs protein Sequence Alignments from deep-Learning of Structural Alignments (SAdLSA, silent ‘d’). The key idea is to implicitly learn the protein folding code from many thousands of structural alignments using experimentally determined protein structures. Results To demonstrate that the folding code was learned, we first show that SAdLSA trained on pure α-helical proteins successfully recognizes pairs of structurally related pure β-sheet protein domains. Subsequent training and benchmarking on larger, highly challenging datasets show significant improvement over established approaches. For challenging cases, SAdLSA is ∼150% better than HHsearch for generating pairwise alignments and ∼50% better for identifying the proteins with the best alignments in a sequence library. The time complexity of SAdLSA is O(N) thanks to GPU acceleration. Availability and implementation Datasets and source codes of SAdLSA are available free of charge for academic users at http://sites.gatech.edu/cssb/sadlsa/. Contact [email protected] or [email protected] Supplementary information Supplementary data are available at Bioinformatics online.

Download Full-text

Deep graph learning of inter-protein contacts

10.1101/2021.08.14.456342 ◽

2021 ◽

Author(s):

Ziwei Xie ◽

Jinbo Xu

Keyword(s):

Deep Learning ◽

Protein Interactions ◽

Language Model ◽

Learning Method ◽

Invariant Representation ◽

Sequence Alignments ◽

Multiple Sequence ◽

Contact Prediction ◽

Graph Learning ◽

Interfacial Contact

Motivation: Inter-protein (interfacial) contact prediction is very useful for in silico structural characterization of protein-protein interactions. Although deep learning has been applied to this problem, its accuracy is not as good as intra-protein contact prediction. Results: We propose a new deep learning method GLINTER (Graph Learning of INTER-protein contacts) for interfacial contact prediction of dimers, leveraging a rotational invariant representation of protein tertiary structures and a pretrained language model of multiple sequence alignments (MSAs). Tested on the 13th and 14th CASP-CAPRI datasets, the average top L/10 precision achieved by GLINTER is 54.35% on the homodimers and 51.56% on all the dimers, much higher than 30.43% obtained by the latest deep learning method DeepHomo on the homodimers and 14.69% obtained by BIPSPI on all the dimers. Our experiments show that GLINTER-predicted contacts help improve selection of docking decoys.

Download Full-text

Protein contact prediction using metagenome sequence data and residual neural networks

Bioinformatics ◽

10.1093/bioinformatics/btz477 ◽

2019 ◽

Vol 36 (1) ◽

pp. 41-48 ◽

Cited By ~ 15

Author(s):

Qi Wu ◽

Zhenling Peng ◽

Ivan Anishchenko ◽

Qian Cong ◽

David Baker ◽

...

Keyword(s):

Neural Networks ◽

Sequence Data ◽

Prediction Method ◽

Supplementary Information ◽

Sequence Profile ◽

Sequence Alignments ◽

Multiple Sequence ◽

Contact Prediction ◽

Benchmark Datasets ◽

Metagenome Sequence

Abstract Motivation Almost all protein residue contact prediction methods rely on the availability of deep multiple sequence alignments (MSAs). However, many proteins from the poorly populated families do not have sufficient number of homologs in the conventional UniProt database. Here we aim to solve this issue by exploring the rich sequence data from the metagenome sequencing projects. Results Based on the improved MSA constructed from the metagenome sequence data, we developed MapPred, a new deep learning-based contact prediction method. MapPred consists of two component methods, DeepMSA and DeepMeta, both trained with the residual neural networks. DeepMSA was inspired by the recent method DeepCov, which was trained on 441 matrices of covariance features. By considering the symmetry of contact map, we reduced the number of matrices to 231, which makes the training more efficient in DeepMSA. Experiments show that DeepMSA outperforms DeepCov by 10–13% in precision. DeepMeta works by combining predicted contacts and other sequence profile features. Experiments on three benchmark datasets suggest that the contribution from the metagenome sequence data is significant with P-values less than 4.04E-17. MapPred is shown to be complementary and comparable the state-of-the-art methods. The success of MapPred is attributed to three factors: the deeper MSA from the metagenome sequence data, improved feature design in DeepMSA and optimized training by the residual neural networks. Availability and implementation http://yanglab.nankai.edu.cn/mappred/. Supplementary information Supplementary data are available at Bioinformatics online.

Download Full-text

PconsC4: fast, accurate and hassle-free contact predictions

Bioinformatics ◽

10.1093/bioinformatics/bty1036 ◽

2018 ◽

Vol 35 (15) ◽

pp. 2677-2679 ◽

Cited By ~ 15

Author(s):

Mirco Michel ◽

David Menéndez Hurtado ◽

Arne Elofsson

Keyword(s):

Deep Learning ◽

State Of The Art ◽

Supplementary Information ◽

Prediction Methods ◽

Coupling Analysis ◽

Contact Prediction ◽

Residue Contact ◽

Direct Coupling Analysis ◽

Computationally Expensive ◽

Contact Predictions

Abstract Motivation Residue contact prediction was revolutionized recently by the introduction of direct coupling analysis (DCA). Further improvements, in particular for small families, have been obtained by the combination of DCA and deep learning methods. However, existing deep learning contact prediction methods often rely on a number of external programs and are therefore computationally expensive. Results Here, we introduce a novel contact predictor, PconsC4, which performs on par with state of the art methods. PconsC4 is heavily optimized, does not use any external programs and therefore is significantly faster and easier to use than other methods. Availability and implementation PconsC4 is freely available under the GPL license from https://github.com/ElofssonLab/PconsC4. Installation is easy using the pip command and works on any system with Python 3.5 or later and a GCC compiler. It does not require a GPU nor special hardware. Supplementary information Supplementary data are available at Bioinformatics online.

Download Full-text

Refined template selection and combination algorithm significantly improves template-based modeling accuracy

Journal of Bioinformatics and Computational Biology ◽

10.1142/s0219720019500069 ◽

2019 ◽

Vol 17 (02) ◽

pp. 1950006 ◽

Cited By ~ 4

Author(s):

Ashish Runthala ◽

Shibasish Chowdhury

Keyword(s):

Structural Information ◽

Protein Structures ◽

Comparative Modeling ◽

Target Sequence ◽

Sequence Alignments ◽

Multiple Sequence ◽

Multiple Sequence Alignments ◽

Modeling Accuracy ◽

Model Protein ◽

Rank And Select

In contrast to ab-initio protein modeling methodologies, comparative modeling is considered as the most popular and reliable algorithm to model protein structure. However, the selection of the best set of templates is still a major challenge. An effective template-ranking algorithm is developed to efficiently select only the reliable hits for predicting the protein structures. The algorithm employs the pairwise as well as multiple sequence alignments of template hits to rank and select the best possible set of templates. It captures several key sequences and structural information of template hits and converts into scores to effectively rank them. This selected set of templates is used to model a target. Modeling accuracy of the algorithm is tested and evaluated on TBM-HA domain containing CASP8, CASP9 and CASP10 targets. On an average, this template ranking and selection algorithm improves GDT-TS, GDT-HA and TM_Score by 3.531, 4.814 and 0.022, respectively. Further, it has been shown that the inclusion of structurally similar templates with ample conformational diversity is crucial for the modeling algorithm to maximally as well as reliably span the target sequence and construct its near-native model. The optimal model sampling also holds the key to predict the best possible target structure.

Download Full-text

Single-sequence protein structure prediction using language models from deep learning

10.1101/2021.08.02.454840 ◽

2021 ◽

Author(s):

Ratul Chowdhury ◽

Nazim Bouatta ◽

Surojit Biswas ◽

Charlotte Rochereau ◽

George M Church ◽

...

Keyword(s):

Deep Learning ◽

Protein Structure ◽

Structure Prediction ◽

Structural Information ◽

Language Model ◽

Learning System ◽

Language Models ◽

Sequence Alignments ◽

Multiple Sequence ◽

Predict Protein Structure

AlphaFold2 and related systems use deep learning to predict protein structure from co-evolutionary relationships encoded in multiple sequence alignments (MSAs). Despite dramatic, recent increases in accuracy, three challenges remain: (i) prediction of orphan and rapidly evolving proteins for which an MSA cannot be generated, (ii) rapid exploration of designed structures, and (iii) understanding the rules governing spontaneous polypeptide folding in solution. Here we report development of an end-to-end differentiable recurrent geometric network (RGN) able to predict protein structure from single protein sequences without use of MSAs. This deep learning system has two novel elements: a protein language model (AminoBERT) that uses a Transformer to learn latent structural information from millions of unaligned proteins and a geometric module that compactly represents Cα backbone geometry. RGN2 outperforms AlphaFold2 and RoseTTAFold (as well as trRosetta) on orphan proteins and is competitive with designed sequences, while achieving up to a billion-fold reduction in compute time. These findings demonstrate the practical and theoretical strengths of protein language models relative to MSAs in structure prediction.

Download Full-text

The evolution of contact prediction: Evidence that contact selection in statistical contact prediction is changing

10.1101/660191 ◽

2019 ◽

Author(s):

Mark Chonofsky ◽

Saulo H. P. de Oliveira ◽

Konrad Krawczyk ◽

Charlotte M. Deane

Keyword(s):

Amino Acids ◽

Protein Structure ◽

Amino Acid ◽

Structure Prediction ◽

Prediction Methods ◽

Sequence Alignments ◽

Multiple Sequence ◽

Contact Prediction ◽

Multiple Sequence Alignments ◽

Physico Chemical

AbstractOver the last few years, the field of protein structure prediction has been transformed by increasingly-accurate contact prediction software. These methods are based on the detection of coevolutionary relationships between residues from multiple sequence alignments. However, despite speculation, there is little evidence of a link between contact prediction and the physico-chemical interactions which drive amino-acid coevolution. Furthermore, existing protocols predict only a fraction of all protein contacts and it is not clear why some contacts are favoured over others.Using a dataset of 863 protein domains, we assessed the physico-chemical interactions of contacts predicted by CCMpred, MetaPSICOV, and DNCON2, as examples of direct coupling analysis, meta-prediction, and deep learning, respectively. To further investigate what sets these predicted contacts apart, we considered correctly-predicted contacts and compared their properties against the protein contacts that were not predicted.We found that predicted contacts tend to form more bonds than non-predicted contacts, which suggests these contacts may be more important. Comparing the contacts predicted by each method, we found that metaPSICOV and DNCON2 favour accuracy whereas CCMPred detects contacts with more bonds. This suggests that the push for higher accuracy may lead to a loss of physico-chemically important contacts.These results underscore the connection between protein physico-chemistry and the coevolutionary couplings that can be derived from multiple sequence alignments. This relationship is likely to be relevant to protein structure prediction and functional analysis of protein structure and may be key to understanding their utility for different problems in structural biology.Author summaryAccurate contact prediction has allowed scientists to predict protein structures with unprecedented levels of accuracy. The success of contact prediction methods, which are based on inferring correlations between amino acids in protein multiple sequence alignments, has prompted a great deal of work to improve the quality of contact prediction, leading to the development of several different methods for detecting amino acids in proximity.In this paper, we investigate the properties of these contact prediction methods. We find that contacts which are predicted differ from the other contacts in the protein, in particular they have more physico-chemical bonds, and the predicted contacts are more strongly conserved than other contacts across protein families. We also compared the properties of different contact prediction methods and found that the characteristics of the predicted sets depend on the prediction method used.Our results point to a link between physico-chemical bonding interactions and the evolutionary history of proteins, a connection which is reflected in their amino acid sequences.

Download Full-text

The evolution of contact prediction: Evidence that contact selection in statistical contact prediction is changing

Bioinformatics ◽

10.1093/bioinformatics/btz816 ◽

2019 ◽

Cited By ~ 2

Author(s):

Mark Chonofsky ◽

Saulo H P de Oliveira ◽

Konrad Krawczyk ◽

Charlotte M Deane

Keyword(s):

Protein Structure ◽

Protein Structure Prediction ◽

Structure Prediction ◽

Supplementary Information ◽

Chemical Interactions ◽

Sequence Alignments ◽

Multiple Sequence ◽

Contact Prediction ◽

Multiple Sequence Alignments ◽

Physico Chemical

Abstract Motivation Over the last few years, the field of protein structure prediction has been transformed by increasingly-accurate contact prediction software. These methods are based on the detection of coevolutionary relationships between residues from multiple sequence alignments. However, despite speculation, there is little evidence of a link between contact prediction and the physico-chemical interactions which drive amino-acid coevolution. Furthermore, existing protocols predict only a fraction of all protein contacts and it is not clear why some contacts are favoured over others. Using a dataset of 863 protein domains, we assessed the physico-chemical interactions of contacts predicted by CCMpred, MetaPSICOV, and DNCON2, as examples of direct coupling analysis, meta-prediction, and deep learning. Results We considered correctly-predicted contacts and compared their properties against the protein contacts that were not predicted. Predicted contacts tend to form more bonds than non-predicted contacts, which suggests these contacts may be more important than contacts that were not predicted. Comparing the contacts predicted by each method, we found that metaPSICOV and DNCON2 favour accuracy whereas CCMPred detects contacts with more bonds. This suggests that the push for higher accuracy may lead to a loss of physico-chemically important contacts. These results underscore the connection between protein physico-chemistry and the coevolutionary couplings that can be derived from multiple sequence alignments. This relationship is likely to be relevant to protein structure prediction and functional analysis of protein structure and may be key to understanding their utility for different problems in structural biology. Availability We use publicly-available databases. Our code is available for download at http://opig.stats.ox.ac.uk/. Supplementary information Supplementary information is available at Bioinformatics online.

Download Full-text