Can three-dimensional contacts in protein structures be predicted by analysis of correlated mutations?

A comparative study of protein structure prediction tools for challenging targets: Snake venom toxins

Protein structure determination is a critical aspect of biological research, enabling us to understand protein function and potential applications. Recent advances in deep learning and artificial intelligence have led to the development of several protein structure prediction tools, such as AlphaFold2 and ColabFold. However, their performance has primarily been evaluated on well-characterised proteins and their ability to predict sturtctures of proteins lacking experimental structures, such as many snake venom toxins, has been less scrutinised. In this study, we evaluated three modelling tools on their prediction of over 1000 snake venom toxin structures for which no experimental structures exist. Our findings show that AlphaFold2 (AF2) performed the best across all assessed parameters. We also observed that ColabFold (CF) only scored slightly worse than AF2, while being computationally less intensive. All tools struggled with regions of intrinsic disorder, such as loops and propeptide regions, and performed well in predicting the structure of functional domains. Overall, our study highlights the importance of exercising caution when working with proteins with no experimental structures available, particularly those that are large and contain flexible regions. Nonetheless, leveraging computational structure prediction tools can provide valuable insights into the modelling of protein interactions with different targets and reveal potential binding sites, active sites, and conformational changes, as well as into the design of potential molecular binders for reagent, diagnostic, or therapeutic purposes.

Continuous evolution of user-defined genes at 1-million-times the genomic mutation rate

When nature maintains or evolves a gene's function over millions of years at scale, it produces a diversity of homologous sequences whose patterns of conservation and change contain rich structural, functional, and historical information about the gene. However, natural gene diversity likely excludes vast regions of functional sequence space and includes phylogenetic and evolutionary eccentricities, limiting what information we can extract. We introduce an accessible experimental approach for compressing long-term gene evolution to laboratory timescales, allowing for the direct observation of extensive adaptation and divergence followed by inference of structural, functional, and environmental constraints for any selectable gene. To enable this approach, we developed a new orthogonal DNA replication (OrthoRep) system that durably hypermutates chosen genes at a rate of >10 -4 substitutions per base in vivo . When OrthoRep was used to evolve a conditionally essential maladapted enzyme, we obtained thousands of unique multi-mutation sequences with many pairs >60 amino acids apart (>15% divergence), revealing known and new factors influencing enzyme adaptation. The fitness of evolved sequences was not predictable by advanced machine learning models trained on natural variation. We suggest that OrthoRep supports the prospective and systematic discovery of constraints shaping gene evolution, uncovering of new regions in fitness landscapes, and general applications in biomolecular engineering.

Co-evolution-based prediction of metal-binding sites in proteomes by machine learning

Metal ions have various important biological roles in proteins, including structural maintenance, molecular recognition and catalysis. Previous methods of predicting metal-binding sites in proteomes were based on either sequence or structural motifs. Here we developed a co-evolution-based pipeline named 'MetalNet' to systematically predict metal-binding sites in proteomes. We applied MetalNet to proteomes of four representative prokaryotic species and predicted 4,849 potential metalloproteins, which substantially expands the currently annotated metalloproteomes. We biochemically and structurally validated previously unannotated metal-binding sites in several proteins, including apo-citrate lyase phosphoribosyl-dephospho-CoA transferase citX, an Escherichia coli enzyme lacking structural or sequence homology to any known metalloprotein (Protein Data Bank (PDB) codes: 7DCM and 7DCN ). MetalNet also successfully recapitulated all known zinc-binding sites from the human spliceosome complex. The pipeline of MetalNet provides a unique and enabling tool for interrogating the hidden metalloproteome and studying metal biology.

Enzyme Substrate Prediction from Three-Dimensional Feature Representations Using Space-Filling Curves

Compact and interpretable structural feature representations are required for accurately predicting properties and function of proteins. In this work, we construct and evaluate three-dimensional feature representations of protein structures based on space-filling curves (SFCs). We focus on the problem of enzyme substrate prediction, using two ubiquitous enzyme families as case studies: the short-chain dehydrogenase/reductases (SDRs) and the S-adenosylmethionine-dependent methyltransferases (SAM-MTases). Space-filling curves such as the Hilbert curve and the Morton curve generate a reversible mapping from discretized three-dimensional to one-dimensional representations and thus help to encode three-dimensional molecular structures in a system-independent way and with only a few adjustable parameters. Using three-dimensional structures of SDRs and SAM-MTases generated using AlphaFold2, we assess the performance of the SFC-based feature representations in predictions on a new benchmark database of enzyme classification tasks including their cofactor and substrate selectivity. Gradient-boosted tree classifiers yield binary prediction accuracy of 0.77-0.91 and area under curve (AUC) characteristics of 0.83-0.92 for the classification tasks. We investigate the effects of amino acid encoding, spatial orientation, and (the few) parameters of SFC-based encodings on the accuracy of the predictions. Our results suggest that geometry-based approaches such as SFCs are promising for generating protein structural representations and are complementary to the existing protein feature representations such as evolutionary scale modeling (ESM) sequence embeddings.

Towards de novo design of transmembrane α-helical assemblies using structural modelling and molecular dynamics simulation

Computational de novo protein design involves iterative processes consisting of amino acid sequence design, structural modelling and scoring, and design validation by synthesis and experimental characterisation. Recent advances in protein structure prediction and modelling methods have enabled the highly efficient and accurate design of water-soluble proteins. However, the design of membrane proteins remains a major challenge. To advance membrane protein design, considering the higher complexity of membrane protein folding, stability, and dynamic interactions between water, ions, lipids, and proteins is an important task. For introducing explicit solvents and membranes to these design methods, all-atom molecular dynamics (MD) simulations of designed proteins provide useful information that cannot be obtained experimentally. In this review, we first describe two major approaches to designing transmembrane α-helical assemblies, consensus and de novo design. We further illustrate recent MD studies of membrane protein folding related to protein design, as well as advanced treatments in molecular models and conformational sampling techniques in the simulations. Finally, we discuss the possibility to introduce MD simulations after the existing static modelling and screening of design decoys as an additional step for refinement of the design, which considers membrane protein folding dynamics and interactions with explicit membranes.

General strategies for using amino acid sequence data to guide biochemical investigation of protein function

The rapid increase of '-omics' data warrants the reconsideration of experimental strategies to investigate general protein function. Studying individual members of a protein family is likely insufficient to provide a complete mechanistic understanding of family functions, especially for diverse families with thousands of known members. Strategies that exploit large amounts of available amino acid sequence data can inspire and guide biochemical experiments, generating broadly applicable insights into a given family. Here we review several methods that utilize abundant sequence data to focus experimental efforts and identify features truly representative of a protein family or domain. First, coevolutionary relationships between residues within primary sequences can be successfully exploited to identify structurally and/or functionally important positions for experimental investigation. Second, functionally important variable residue positions typically occupy a limited sequence space, a property useful for guiding biochemical characterization of the effects of the most physiologically and evolutionarily relevant amino acids. Third, amino acid sequence variation within domains shared between different protein families can be used to sort a particular domain into multiple subtypes, inspiring further experimental designs. Although generally applicable to any kind of protein domain because they depend solely on amino acid sequences, the second and third approaches are reviewed in detail because they appear to have been used infrequently and offer immediate opportunities for new advances. Finally, we speculate that future technologies capable of analyzing and manipulating conserved and variable aspects of the three-dimensional structures of a protein family could lead to broad insights not attainable by current methods.

Combining cysteine scanning with chemical labeling to map protein-protein interactions and infer bound structure in an intrinsically disordered region

The Mycobacterium tuberculosis genome harbours nine toxin-antitoxin (TA) systems of the mazEF family. These consist of two proteins, a toxin and an antitoxin, encoded in an operon. While the toxin has a conserved fold, the antitoxins are structurally diverse and the toxin binding region is typically intrinsically disordered before binding. We describe high throughput methodology for accurate mapping of interfacial residues and apply it to three MazEF complexes. The method involves screening one partner protein against a panel of chemically masked single cysteine mutants of its interacting partner, displayed on the surface of yeast cells. Such libraries have much lower diversity than those generated by saturation mutagenesis, simplifying library generation and data analysis. Further, because of the steric bulk of the masking reagent, labeling of virtually all exposed epitope residues should result in loss of binding, and buried residues are inaccessible to the labeling reagent. The binding residues are deciphered by probing the loss of binding to the labeled cognate partner by flow cytometry. Using this methodology, we have identified the interfacial residues for MazEF3, MazEF6 and MazEF9 TA systems of M. tuberculosis. In the case of MazEF9, where a crystal structure was available, there was excellent agreement between our predictions and the crystal structure, superior to those with AlphaFold2. We also report detailed biophysical characterization of the MazEF3 and MazEF9 TA systems and measured the relative affinities between cognate and non-cognate toxin–antitoxin partners in order to probe possible cross-talk between these systems.

Mechanistic insights into global suppressors of protein folding defects

Most amino acid substitutions in a protein either lead to partial loss-of-function or are near neutral. Several studies have shown the existence of second-site mutations that can rescue defects caused by diverse loss-of-function mutations. Such global suppressor mutations are key drivers of protein evolution. However, the mechanisms responsible for such suppression remain poorly understood. To address this, we characterized multiple suppressor mutations both in isolation and in combination with inactive mutants. We examined six global suppressors of the bacterial toxin CcdB, the known M182T global suppressor of TEM-1 β-lactamase, the N239Y global suppressor of p53-DBD and three suppressors of the SARS-CoV-2 spike Receptor Binding Domain. When coupled to inactive mutants, they promote increased in-vivo solubilities as well as regain-of-function phenotypes. In the case of CcdB, where novel suppressors were isolated, we determined the crystal structures of three such suppressors to obtain insight into the specific molecular interactions responsible for the observed effects. While most individual suppressors result in small stability enhancements relative to wildtype, which can be combined to yield significant stability increments, thermodynamic stabilisation is neither necessary nor sufficient for suppressor action. Instead, in diverse systems, we observe that individual global suppressors greatly enhance the foldability of buried site mutants, primarily through increase in refolding rate parameters measured in vitro. In the crowded intracellular environment, mutations that slow down folding likely facilitate off-pathway aggregation. We suggest that suppressor mutations that accelerate refolding can counteract this, enhancing the yield of properly folded, functional protein in vivo.

Web-Based Protein Interactions Calculator Identifies Likely Proteome Coevolution with Alzheimer’s Disease-Associated Proteins

Protein–protein functional interactions arise from either transitory or permanent biomolecular associations and often lead to the coevolution of the interacting residues. Although mutual information has traditionally been used to identify coevolving residues within the same protein, its application between coevolving proteins remains largely uncharacterized. Therefore, we developed the Protein Interactions Calculator (PIC) to efficiently identify coevolving residues between two protein sequences using mutual information. We verified the algorithm using 2102 known human protein interactions and 233 known bacterial protein interactions, with a respective 1975 and 252 non-interacting protein controls. The average PIC score for known human protein interactions was 4.5 times higher than non-interacting proteins (p = 1.03 × 10⁻¹⁰⁸) and 1.94 times higher in bacteria (p = 1.22 × 10⁻³⁵). We then used the PIC scores to determine the probability that two proteins interact. Using those probabilities, we paired 37 Alzheimer’s disease-associated proteins with 8608 other proteins and determined the likelihood that each pair interacts, which we report through a web interface. The PIC had significantly higher sensitivity and residue-specific resolution not available in other algorithms. Therefore, we propose that the PIC can be used to prioritize potential protein interactions, which can lead to a better understanding of biological processes and additional therapeutic targets belonging to protein interaction groups.

From systems to structure - using genetic data to model protein structures

Understanding the effects of genetic variation is a fundamental problem in biology that requires methods to analyse both physical and functional consequences of sequence changes at systems-wide and mechanistic scales. To achieve a systems view, protein interaction networks map which proteins physically interact, while genetic interaction networks inform on the phenotypic consequences of perturbing these protein interactions. Until recently, understanding the molecular mechanisms that underlie these interactions often required biophysical methods to determine the structures of the proteins involved. The past decade has seen the emergence of new approaches based on coevolution, deep mutational scanning and genome-scale genetic or chemical-genetic interaction mapping that enable modelling of the structures of individual proteins or protein complexes. Here, we review the emerging use of large-scale genetic datasets and deep learning approaches to model protein structures and their interactions, and discuss the integration of structural data from different sources.

Molecular evolution and structural analyses of the spike glycoprotein from Brazilian SARS-CoV-2 genomes: the impact of selected mutations

The COVID-19 pandemic caused by SARS-CoV-2 has reached by February 2022 more than 380 million cases and 5.5 million deaths worldwide since its beginning in late 2019, leading to enhanced concern in the scientific community and the general population. One of the most important pieces of this host-pathogen interaction is the spike protein, which binds to the hACE2 cell receptor, mediates the membrane fusion and is the major target of neutralizing antibodies against SARS-CoV-2. The multiple amino acid substitutions observed in this region, specially in RBD have enhanced the hACE2 binding affinity and led to several modifications in the mechanisms of SARS-CoV-2 pathogenesis, improving the viral fitness and/or promoting immune evasion, with potential impact in the vaccine development. In this work, we identified 48 sites under selective pressures, 17 of them with the strongest evidence by the HyPhy tests, including VOC related mutation sites 138, 142, 222, 262, 484, 681, and 845, among others. The coevolutionary analysis identified 28 sites found not to be conditionally independent, such as E484K-N501Y. The molecular dynamics and free energy estimates showed the structural stabilizing effect and the higher impact of E484K for enhanced binding affinity between the spike RBD and hACE2 in P.1 and P.2 lineages (specially with L452V). Structural changes were also identified in the hACE molecule when interacting with B.1.1.7 RDB. Despite some destabilizing substitutions, a stabilizing effect was identified for the majority of the positively selected mutations.Communicated by Ramaswamy H. Sarma.

Boltzmann Machine Learning and Regularization Methods for Inferring Evolutionary Fields and Couplings From a Multiple Sequence Alignment

The inverse Potts problem to infer a Boltzmann distribution for homologous protein sequences from their single-site and pairwise amino acid frequencies recently attracts a great deal of attention in the studies of protein structure and evolution. We study regularization and learning methods and how to tune regularization parameters to correctly infer interactions in Boltzmann machine learning. Using L₂ regularization for fields, group L₁ for couplings is shown to be very effective for sparse couplings in comparison with L₂ and L₁. Two regularization parameters are tuned to yield equal values for both the sample and ensemble averages of evolutionary energy. Both averages smoothly change and converge, but their learning profiles are very different between learning methods. The Adam method is modified to make stepsize proportional to the gradient for sparse couplings and to use a soft-thresholding function for group L₁. It is shown by first inferring interactions from protein sequences and then from Monte Carlo samples that the fields and couplings can be well recovered, but that recovering the pairwise correlations in the resolution of a total energy is harder for the natural proteins than for the protein-like sequences. Selective temperature for folding/structural constrains in protein evolution is also estimated.

A minimal yet flexible likelihood framework to assess correlated evolution

An evolutionary process is reflected in the sequence of changes of any trait (e.g. morphological or molecular) through time. Yet, a better understanding of evolution would be procured by characterizing correlated evolution, or when two or more evolutionary processes interact. Previously developed parametric methods often require significant computing time as they rely on the estimation of many parameters. Here we propose a minimal likelihood framework modelling the joint evolution of two traits on a known phylogenetic tree. The type and strength of correlated evolution is characterized by a few parameters tuning mutation rates of each trait and interdependencies between these rates. The framework can be applied to study any discrete trait or character ranging from nucleotide substitution to gain or loss of a biological function. More specifically, it can be used to 1) test for independence between two evolutionary processes, 2) identify the type of interaction between them and 3) estimate parameter values of the most likely model of interaction. In the current implementation, the method takes as input a phylogenetic tree with discrete evolutionary events mapped on its branches. The method then maximizes the likelihood for one or several chosen scenarios. The strengths and limits of the method, as well as its relative power compared to a few other methods, are assessed using both simulations and data from 16S rRNA sequences in a sample of 54 γ-enterobacteria. We show that, even with datasets of fewer than 100 species, the method performs well in parameter estimation and in evolutionary model selection.

Conservation and coevolution determine evolvability of different classes of disordered residues in human intrinsically disordered proteins

Structure, function, and evolution are interdependent properties of proteins. Diversity of protein functions arising from structural variations is a potential driving force behind protein evolvability. Intrinsically disordered proteins or regions (IDPs or IDRs) lack well-defined structure under normal physiological conditions, yet, they are highly functional. Increased occurrence of IDPs in eukaryotes compared to prokaryotes indicates strong correlation of protein evolution and disorderedness. IDPs generally have higher evolution rate compared to globular proteins. Structural pliability allows IDPs to accommodate multiple mutations without affecting their functional potential. Nevertheless, how evolutionary signals vary between different classes of disordered residues (DRs) in IDPs is poorly understood. This study addresses variation of evolutionary behavior in terms of residue conservation and intra-protein coevolution among structural and functional classes of DRs in IDPs. Analyses are performed on 579 human IDPs, which are classified based on length of IDRs, interacting partners and functional classes. We find short IDRs are less conserved than long IDRs or full IDPs. Functional classes which require flexibility and specificity to perform their activity comparatively evolve slower than others. Disorder promoting amino acids evolve faster than order promoting amino acids. Pro, Gly, Ile, and Phe have unique coevolving nature which further emphasizes on their roles in IDPs. This study sheds light on evolutionary footprints in different classes of DRs from human IDPs and enhances our understanding of the structural and functional potential of IDPs.

Extending the Horizon of Homology Detection with Coevolution-based Structure Prediction

Traditional sequence analysis algorithms fail to identify distant homologies when they lie beyond a detection horizon. In this review, we discuss how co-evolution-based contact and distance prediction methods are pushing back this homology detection horizon, thereby yielding new functional insights and experimentally testable hypotheses. Based on correlated substitutions, these methods divine three-dimensional constraints among amino acids in protein sequences that were previously devoid of all annotated domains and repeats. The new algorithms discern hidden structure in an otherwise featureless sequence landscape. Their revelatory impact promises to be as profound as the use, by archaeologists, of ground-penetrating radar to discern long-hidden, subterranean structures. As examples of this, we describe how triplicated structures reflecting longin domains in MON1A-like proteins, or UVR-like repeats in DISC1, emerge from their predicted contact and distance maps. These methods also help to resolve structures that do not conform to a "beads-on-a-string" model of protein domains. In one such example, we describe CFAP298 whose ubiquitin-like domain was previously challenging to perceive owing to a large sequence insertion within it. More generally, the new algorithms permit an easier appreciation of domain families and folds whose evolution involved structural insertion or rearrangement. As we exemplify with α1-antitrypsin, coevolution-based predicted contacts may also yield insights into protein dynamics and conformational change. This new combination of structure prediction (using innovative co-evolution based methods) and homology inference (using more traditional sequence analysis approaches) shows great promise for bringing into view a sea of evolutionary relationships that had hitherto lain far beyond the horizon of homology detection.

AlphaFold and Implications for Intrinsically Disordered Proteins

Accurate predictions of the three-dimensional structures of proteins from their amino acid sequences have come of age. AlphaFold, a deep learning-based approach to protein structure prediction, shows remarkable success in independent assessments of prediction accuracy. A significant epoch in structural bioinformatics was the structural annotation of over 98% of protein sequences in the human proteome. Interestingly, many predictions feature regions of very low confidence, and these regions largely overlap with intrinsically disordered regions (IDRs). That over 30% of regions within the proteome are disordered is congruent with estimates that have been made over the past two decades, as intense efforts have been undertaken to generalize the structure-function paradigm to include the importance of conformational heterogeneity and dynamics. With structural annotations from AlphaFold in hand, there is the temptation to draw inferences regarding the "structures" of IDRs and their interactomes. Here, we offer a cautionary note regarding the misinterpretations that might ensue and highlight efforts that provide concrete understanding of sequence-ensemble-function relationships of IDRs. This perspective is intended to emphasize the importance of IDRs in sequence-function relationships (SERs) and to highlight how one might go about extracting quantitative SERs to make sense of how IDRs function.

Highly accurate protein structure prediction with AlphaFold

Proteins are essential to life, and understanding their structure can facilitate a mechanistic understanding of their function. Through an enormous experimental effort1-4, the structures of around 100,000 unique proteins have been determined5, but this represents a small fraction of the billions of known protein sequences6,7. Structural coverage is bottlenecked by the months to years of painstaking effort required to determine a single protein structure. Accurate computational approaches are needed to address this gap and to enable large-scale structural bioinformatics. Predicting the three-dimensional structure that a protein will adopt based solely on its amino acid sequence-the structure prediction component of the 'protein folding problem'8-has been an important open research problem for more than 50 years9. Despite recent progress10-14, existing methods fall far short of atomic accuracy, especially when no homologous structure is available. Here we provide the first computational method that can regularly predict protein structures with atomic accuracy even in cases in which no similar structure is known. We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Critical Assessment of protein Structure Prediction (CASP14)15, demonstrating accuracy competitive with experimental structures in a majority of cases and greatly outperforming other methods. Underpinning the latest version of AlphaFold is a novel machine learning approach that incorporates physical and biological knowledge about protein structure, leveraging multi-sequence alignments, into the design of the deep learning algorithm.

Experimentally-driven protein structure modeling

Revolutions in natural and exact sciences started at the dawn of last century have led to the explosion of theoretical, experimental, and computational approaches to determine structures of molecules, complexes, as well as their rich conformational dynamics. Since different experimental methods produce information that is attributed to specific time and length scales, corresponding computational methods have to be tailored to these scales and experiments. These methods can be then combined and integrated in scales, hence producing a fuller picture of molecular structure and motion from the "puzzle pieces" offered by various experiments. Here, we describe a number of computational approaches to utilize experimental data to glance into structure of proteins and understand their dynamics. We will also discuss the limitations and the resolution of the constraints-based modeling approaches. SIGNIFICANCE: Experimentally-driven computational structure modeling and determination is a rapidly evolving alternative to traditional approaches for molecular structure determination. These new hybrid experimental-computational approaches are proving to be a powerful microscope to glance into the structural features of intrinsically or partially disordered proteins, dynamics of molecules and complexes. In this review, we describe various approaches in the field of experimentally-driven computational structure modeling.

Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data

The fast replication rate and lack of repair mechanisms of human immunodeficiency virus (HIV) contribute to its high mutation frequency, with some mutations resulting in the evolution of resistance to antiretroviral therapies (ART). As such, studying HIV drug resistance allows for real-time evaluation of evolutionary mechanisms. Characterizing the biological process of drug resistance is also critically important for sustained effectiveness of ART. Investigating the link between "black box" deep learning methods applied to this problem and evolutionary principles governing drug resistance has been overlooked to date. Here, we utilized publicly available HIV-1 sequence data and drug resistance assay results for 18 ART drugs to evaluate the performance of three architectures (multilayer perceptron, bidirectional recurrent neural network, and convolutional neural network) for drug resistance prediction, jointly with biological analysis. We identified convolutional neural networks as the best performing architecture and displayed a correspondence between the importance of biologically relevant features in the classifier and overall performance. Our results suggest that the high classification performance of deep learning models is indeed dependent on drug resistance mutations (DRMs). These models heavily weighted several features that are not known DRM locations, indicating the utility of model interpretability to address causal relationships in viral genotype-phenotype data.

Secondary Structure and Contact Guided Differential Evolution for Protein Structure Prediction

Ab initio protein tertiary structure prediction is one of the long-standing problems in structural bioinformatics. With the help of residue-residue contact and secondary structure prediction information, the accuracy of ab initio structure prediction can be enhanced. In this study, an improved differential evolution with secondary structure and residue-residue contact information referred to as SCDE is proposed for protein structure prediction. In SCDE, two score models based on secondary structure and contact information are proposed, and two selection strategies, namely, secondary structure-based selection strategy and contact-based selection strategy, are designed to guide conformation space search. A probability distribution function is designed to balance these two selection strategies. Experimental results on a benchmark dataset with 28 proteins and four free model targets in CASP12 demonstrate that the proposed SCDE is effective and efficient.

Ensembling multiple raw coevolutionary features with deep residual neural networks for contact-map prediction in CASP13

We report the results of residue-residue contact prediction of a new pipeline built purely on the learning of coevolutionary features in the CASP13 experiment. For a query sequence, the pipeline starts with the collection of multiple sequence alignments (MSAs) from multiple genome and metagenome sequence databases using two complementary Hidden Markov Model (HMM)-based searching tools. Three profile matrices, built on covariance, precision, and pseudolikelihood maximization respectively, are then created from the MSAs, which are used as the input features of a deep residual convolutional neural network architecture for contact-map training and prediction. Two ensembling strategies have been proposed to integrate the matrix features through end-to-end training and stacking, resulting in two complementary programs called TripletRes and ResTriplet, respectively. For the 31 free-modeling domains that do not have homologous templates in the PDB, TripletRes and ResTriplet generated comparable results with an average accuracy of 0.640 and 0.646, respectively, for the top L/5 long-range predictions, where 71% and 74% of the cases have an accuracy above 0.5. Detailed data analyses showed that the strength of the pipeline is due to the sensitive MSA construction and the advanced strategies for coevolutionary feature ensembling. Domain splitting was also found to help enhance the contact prediction performance. Nevertheless, contact models for tail regions, which often involve a high number of alignment gaps, and for targets with few homologous sequences are still suboptimal. Development of new approaches where the model is specifically trained on these regions and targets might help address these problems.

Evolutionary coupling saturation mutagenesis: Coevolution-guided identification of distant sites influencing Bacillus naganoensis pullulanase activity

Pullulanases are well-known debranching enzymes hydrolyzing α-1,6-glycosidic linkages. To date, engineering of pullulanase is mainly focused on catalytic pocket or domain tailoring based on structure/sequence information. Saturation mutagenesis-involved directed evolution is, however, limited by the low number of mutational sites compatible with combinatorial libraries of feasible size. Using Bacillus naganoensis pullulanase as a target protein, here we introduce the 'evolutionary coupling saturation mutagenesis' (ECSM) approach: residue pair covariances are calculated to identify residues for saturation mutagenesis, focusing directed evolution on residue pairs playing important roles in natural evolution. Evolutionary coupling (EC) analysis identified seven residue pairs as evolutionary mutational hotspots. Subsequent saturation mutagenesis yielded variants with enhanced catalytic activity. The functional pairs apparently represent distant sites affecting enzyme activity.

Predicting protein inter-residue contacts using composite likelihood maximization and deep learning

BACKGROUND

Accurate prediction of inter-residue contacts of a protein is important to calculating its tertiary structure. Analysis of co-evolutionary events among residues has been proved effective in inferring inter-residue contacts. The Markov random field (MRF) technique, although being widely used for contact prediction, suffers from the following dilemma: the actual likelihood function of MRF is accurate but time-consuming to calculate; in contrast, approximations to the actual likelihood, say pseudo-likelihood, are efficient to calculate but inaccurate. Thus, how to achieve both accuracy and efficiency simultaneously remains a challenge.

RESULTS

In this study, we present such an approach (called clmDCA) for contact prediction. Unlike plmDCA using pseudo-likelihood, i.e., the product of conditional probability of individual residues, our approach uses composite-likelihood, i.e., the product of conditional probability of all residue pairs. Composite likelihood has been theoretically proved as a better approximation to the actual likelihood function than pseudo-likelihood. Meanwhile, composite likelihood is still efficient to maximize, thus ensuring the efficiency of clmDCA. We present comprehensive experiments on popular benchmark datasets, including PSICOV dataset and CASP-11 dataset, to show that: i) clmDCA alone outperforms the existing MRF-based approaches in prediction accuracy. ii) When equipped with deep learning technique for refinement, the prediction accuracy of clmDCA was further significantly improved, suggesting the suitability of clmDCA for subsequent refinement procedure. We further present a successful application of the predicted contacts to accurately build tertiary structures for proteins in the PSICOV dataset.

CONCLUSIONS

Composite likelihood maximization algorithm can efficiently estimate the parameters of Markov Random Fields and can improve the prediction accuracy of protein inter-residue contacts.

Assessing the accuracy of contact predictions in CASP13

The accuracy of sequence-based tertiary contact predictions was assessed in a blind prediction experiment at the CASP13 meeting. After 4 years of significant improvements in prediction accuracy, another dramatic advance has taken place since CASP12 was held 2 years ago. The precision of predicting the top L/5 contacts in the free modeling category, where L is the corresponding length of the protein in residues, has exceeded 70%. As a comparison, the best-performing group at CASP12 with a 47% precision would have finished below the top 1/3 of the CASP13 groups. Extensively trained deep neural network approaches dominate the top performing algorithms, which appear to efficiently integrate information on coevolving residues and interacting fragments or possibly utilize memories of sequence similarities and sometimes can deliver accurate results even in the absence of virtually any target specific evolutionary information. If the current performance is evaluated by F-score on L contacts, it stands around 24% right now, which, despite the tremendous impact and advance in improving its utility for structure modeling, also suggests that there is much room left for further improvement.

Bio-knowledge-based filters improve residue-residue contact prediction accuracy

Motivation

Residue-residue contact prediction through direct coupling analysis has reached impressive accuracy, but yet higher accuracy will be needed to allow for routine modelling of protein structures. One way to improve the prediction accuracy is to filter predicted contacts using knowledge about the particular protein of interest or knowledge about protein structures in general.

Results

We focus on the latter and discuss a set of filters that can be used to remove false positive contact predictions. Each filter depends on one or a few cut-off parameters for which the filter performance was investigated. Combining all filters while using default parameters resulted for a test set of 851 protein domains in the removal of 29% of the predictions of which 92% were indeed false positives.

Availability and implementation

All data and scripts are available at http://comprec-lin.iiar.pwr.edu.pl/FPfilter/.

Supplementary information

Supplementary data are available at Bioinformatics online.

Evolutionary couplings detect side-chain interactions

Patterns of amino acid covariation in large protein sequence alignments can inform the prediction of de novo protein structures, binding interfaces, and mutational effects. While algorithms that detect these so-called evolutionary couplings between residues have proven useful for practical applications, less is known about how and why these methods perform so well, and what insights into biological processes can be gained from their application. Evolutionary coupling algorithms are commonly benchmarked by comparison to true structural contacts derived from solved protein structures. However, the methods used to determine true structural contacts are not standardized and different definitions of structural contacts may have important consequences for interpreting the results from evolutionary coupling analyses and understanding their overall utility. Here, we show that evolutionary coupling analyses are significantly more likely to identify structural contacts between side-chain atoms than between backbone atoms. We use both simulations and empirical analyses to highlight that purely backbone-based definitions of true residue–residue contacts (i.e., based on the distance between Cα atoms) may underestimate the accuracy of evolutionary coupling algorithms by as much as 40% and that a commonly used reference point (Cβ atoms) underestimates the accuracy by 10–15%. These findings show that co-evolutionary outcomes differ according to which atoms participate in residue–residue interactions and suggest that accounting for different interaction types may lead to further improvements to contact-prediction methods.

Coevolutionary Landscape of Kinase Family Proteins: Sequence Probabilities and Functional Motifs

The protein kinase catalytic domain is one of the most abundant domains across all branches of life. Although kinases share a common core function of phosphoryl-transfer, they also have wide functional diversity and play varied roles in cell signaling networks, and for this reason are implicated in a number of human diseases. This functional diversity is primarily achieved through sequence variation, and uncovering the sequence-function relationships for the kinase family is a major challenge. In this study we use a statistical inference technique inspired by statistical physics, which builds a coevolutionary "Potts" Hamiltonian model of sequence variation in a protein family. We show how this model has sufficient power to predict the probability of specific subsequences in the highly diverged kinase family, which we verify by comparing the model's predictions with experimental observations in the Uniprot database. We show that the pairwise (residue-residue) interaction terms of the statistical model are necessary and sufficient to capture higher-than-pairwise mutation patterns of natural kinase sequences. We observe that previously identified functional sets of residues have much stronger correlated interaction scores than are typical.

Detecting Amino Acid Coevolution with Bayesian Graphical Models

The comparative study of homologous proteins can provide abundant information about the functional and structural constraints on protein evolution. For example, an amino acid substitution that is deleterious may become permissive in the presence of another substitution at a second site of the protein. A popular approach for detecting coevolving residues is by looking for correlated substitution events on branches of the molecular phylogeny relating the protein-coding sequences. Here we describe a machine learning method (Bayesian graphical models) implemented in the open-source phylogenetic software package HyPhy, http://hyphy.org , for extracting a network of coevolving residues from a sequence alignment.

Numerical Encodings of Amino Acids in Multivariate Gaussian Modeling of Protein Multiple Sequence Alignments

Residues in proteins that are in close spatial proximity are more prone to covariate as their interactions are likely to be preserved due to structural and evolutionary constraints. If we can detect and quantify such covariation, physical contacts may then be predicted in the structure of a protein solely from the sequences that decorate it. To carry out such predictions, and following the work of others, we have implemented a multivariate Gaussian model to analyze correlation in multiple sequence alignments. We have explored and tested several numerical encodings of amino acids within this model. We have shown that 1D encodings based on amino acid biochemical and biophysical properties, as well as higher dimensional encodings computed from the principal components of experimentally derived mutation/substitution matrices, do not perform as well as a simple twenty dimensional encoding with each amino acid represented with a vector of one along its own dimension and zero elsewhere. The optimum obtained from representations based on substitution matrices is reached by using 10 to 12 principal components; the corresponding performance is less than the performance obtained with the 20-dimensional binary encoding. We highlight also the importance of the prior when constructing the multivariate Gaussian model of a multiple sequence alignment.

Combining Evolutionary Covariance and NMR Data for Protein Structure Determination

Accurate protein structure determination by solution-state NMR is challenging for proteins greater than about 20kDa, for which extensive perdeuteration is generally required, providing experimental data that are incomplete (sparse) and ambiguous. However, the massive increase in evolutionary sequence information coupled with advances in methods for sequence covariance analysis can provide reliable residue-residue contact information for a protein from sequence data alone. These "evolutionary couplings (ECs)" can be combined with sparse NMR data to determine accurate 3D protein structures. This hybrid "EC-NMR" method has been developed using NMR data for several soluble proteins and validated by comparison with corresponding reference structures determined by X-ray crystallography and/or conventional NMR methods. For small proteins, only backbone resonance assignments are utilized, while for larger proteins both backbone and some sidechain methyl resonance assignments are generally required. ECs can be combined with sparse NMR data obtained on deuterated, selectively protonated protein samples to provide structures that are more accurate and complete than those obtained using such sparse NMR data alone. EC-NMR also has significant potential for analysis of protein structures from solid-state NMR data and for studies of integral membrane proteins. The requirement that ECs are consistent with NMR data recorded on a specific member of a protein family, under specific conditions, also allows identification of ECs that reflect alternative allosteric or excited states of the protein structure.

Synthetic protein alignments by CCMgen quantify noise in residue-residue contact prediction

Compensatory mutations between protein residues in physical contact can manifest themselves as statistical couplings between the corresponding columns in a multiple sequence alignment (MSA) of the protein family. Conversely, large coupling coefficients predict residue contacts. Methods for de-novo protein structure prediction based on this approach are becoming increasingly reliable. Their main limitation is the strong systematic and statistical noise in the estimation of coupling coefficients, which has so far limited their application to very large protein families. While most research has focused on improving predictions by adding external information, little progress has been made to improve the statistical procedure at the core, because our lack of understanding of the sources of noise poses a major obstacle. First, we show theoretically that the expectation value of the coupling score assuming no coupling is proportional to the product of the square roots of the column entropies, and we propose a simple entropy bias correction (EntC) that subtracts out this expectation value. Second, we show that the average product correction (APC) includes the correction of the entropy bias, partly explaining its success. Third, we have developed CCMgen, the first method for simulating protein evolution and generating realistic synthetic MSAs with pairwise statistical residue couplings. Fourth, to learn exact statistical models that reliably reproduce observed alignment statistics, we developed CCMpredPy, an implementation of the persistent contrastive divergence (PCD) method for exact inference. Fifth, we demonstrate how CCMgen and CCMpredPy can facilitate the development of contact prediction methods by analysing the systematic noise contributions from phylogeny and entropy. Using the entropy bias correction, we can disentangle both sources of noise and find that entropy contributes roughly twice as much noise as phylogeny.

Knowledge about the three-dimensional structure of proteins is key to understanding their function and role in biological processes and diseases. The experimental structure determination techniques, such as X-ray crystallography or electron cryo-microscopy, are labour intensive, time-consuming and expensive. Therefore, complementary computational methods to predict a protein’s structure have become indispensable. Over the last years, immense progress has been made in predicting protein structures from their amino acid sequence by utilizing highly accurate predictions of spatial contacts between amino acid residues as constraints in folding simulations. However, contact prediction methods require large numbers of homologous protein sequences in order to discriminate between signal and noise. A major obstacle preventing progress on the statistical methodology is our limited understanding of the different components of noise that are known to affect the predictions. We provide two tools, CCMpredPy and CCMgen, that can be used to learn highly accurate statistical models for contact prediction and to simulate protein evolution according to the statistical constraints between positions of residues as specified by these models, respectively. We showcase their usefulness by quantifying the relative contribution of noise arising from entropy and phylogeny on the predicted contacts, which will facilitate the improvement of the statistical methodology.

Combined approaches from physics, statistics, and computer science for ab initio protein structure prediction: ex unitate vires (unity is strength)?

Connecting the dots among the amino acid sequence of a protein, its structure, and its function remains a central theme in molecular biology, as it would have many applications in the treatment of illnesses related to misfolding or protein instability. As a result of high-throughput sequencing methods, biologists currently live in a protein sequence-rich world. However, our knowledge of protein structure based on experimental data remains comparatively limited. As a consequence, protein structure prediction has established itself as a very active field of research to fill in this gap. This field, once thought to be reserved for theoretical biophysicists, is constantly reinventing itself, borrowing ideas informed by an ever-increasing assembly of scientific domains, from biology, chemistry, (statistical) physics, mathematics, computer science, statistics, bioinformatics, and more recently data sciences. We review the recent progress arising from this integration of knowledge, from the development of specific computer architecture to allow for longer timescales in physics-based simulations of protein folding to the recent advances in predicting contacts in proteins based on detection of coevolution using very large data sets of aligned protein sequences.

Deep Learning in Biomedical Data Science

Since the 1980s, deep learning and biomedical data have been coevolving and feeding each other. The breadth, complexity, and rapidly expanding size of biomedical data have stimulated the development of novel deep learning methods, and application of these methods to biomedical data have led to scientific discoveries and practical solutions. This overview provides technical and historical pointers to the field, and surveys current applications of deep learning to biomedical data organized around five subareas, roughly of increasing spatial scale: chemoinformatics, proteomics, genomics and transcriptomics, biomedical imaging, and health care. The black box problem of deep learning methods is also briefly discussed.

Contact prediction is hardest for the most informative contacts, but improves with the incorporation of contact potentials

Co-evolution between pairs of residues in a multiple sequence alignment (MSA) of homologous proteins has long been proposed as an indicator of structural contacts. Recently, several methods, such as direct-coupling analysis (DCA) and MetaPSICOV, have been shown to achieve impressive rates of contact prediction by taking advantage of considerable sequence data. In this paper, we show that prediction success rates are highly sensitive to the structural definition of a contact, with more permissive definitions (i.e., those classifying more pairs as true contacts) naturally leading to higher positive predictive rates, but at the expense of the amount of structural information contributed by each contact. Thus, the remaining limitations of contact prediction algorithms are most noticeable in conjunction with geometrically restrictive contacts—precisely those that contribute more information in structure prediction. We suggest that to improve prediction rates for such “informative” contacts one could combine co-evolution scores with additional indicators of contact likelihood. Specifically, we find that when a pair of co-varying positions in an MSA is occupied by residue pairs with favorable statistical contact energies, that pair is more likely to represent a true contact. We show that combining a contact potential metric with DCA or MetaPSICOV performs considerably better than DCA or MetaPSICOV alone, respectively. This is true regardless of contact definition, but especially true for stricter and more informative contact definitions. In summary, this work outlines some remaining challenges to be addressed in contact prediction and proposes and validates a promising direction towards improvement.

NeBcon: protein contact map prediction using neural network training coupled with naïve Bayes classifiers

Motivation

Recent CASP experiments have witnessed exciting progress on folding large-size non-humongous proteins with the assistance of co-evolution based contact predictions. The success is however anecdotal due to the requirement of the contact prediction methods for the high volume of sequence homologs that are not available to most of the non-humongous protein targets. Development of efficient methods that can generate balanced and reliable contact maps for different type of protein targets is essential to enhance the success rate of the ab initio protein structure prediction.

Results

We developed a new pipeline, NeBcon, which uses the naïve Bayes classifier (NBC) theorem to combine eight state of the art contact methods that are built from co-evolution and machine learning approaches. The posterior probabilities of the NBC model are then trained with intrinsic structural features through neural network learning for the final contact map prediction. NeBcon was tested on 98 non-redundant proteins, which improves the accuracy of the best co-evolution based meta-server predictor by 22%; the magnitude of the improvement increases to 45% for the hard targets that lack sequence and structural homologs in the databases. Detailed data analysis showed that the major contribution to the improvement is due to the optimized NBC combination of the complementary information from both co-evolution and machine learning predictions. The neural network training also helps to improve the coupling of the NBC posterior probability and the intrinsic structural features, which were found particularly important for the proteins that do not have sufficient number of homologous sequences to derive reliable co-evolution profiles.

Availiablity and Implementation

On-line server and standalone package of the program are available at http://zhanglab.ccmb.med.umich.edu/NeBcon/ .

Contact

zhng@umich.edu.

Supplementary information

Supplementary data are available at Bioinformatics online.

Assessment of contact predictions in CASP12: Co-evolution and deep learning coming of age

Following up on the encouraging results of residue-residue contact prediction in the CASP11 experiment, we present the analysis of predictions submitted for CASP12. The submissions include predictions of 34 groups for 38 domains classified as free modeling targets which are not accessible to homology-based modeling due to a lack of structural templates. CASP11 saw a rise of coevolution-based methods outperforming other approaches. The improvement of these methods coupled to machine learning and sequence database growth are most likely the main driver for a significant improvement in average precision from 27% in CASP11 to 47% in CASP12. In more than half of the targets, especially those with many homologous sequences accessible, precisions above 90% were achieved with the best predictors reaching a precision of 100% in some cases. We furthermore tested the impact of using these contacts as restraints in ab initio modeling of 14 single-domain free modeling targets using Rosetta. Adding contacts to the Rosetta calculations resulted in improvements of up to 26% in GDT_TS within the top five structures.

Power law tails in phylogenetic systems

Covariance analysis of protein sequence alignments uses coevolving pairs of sequence positions to predict features of protein structure and function. However, current methods ignore the phylogenetic relationships between sequences, potentially corrupting the identification of covarying positions. Here, we use random matrix theory to demonstrate the existence of a power law tail that distinguishes the spectrum of covariance caused by phylogeny from that caused by structural interactions. The power law is essentially independent of the phylogenetic tree topology, depending on just two parameters-the sequence length and the average branch length. We demonstrate that these power law tails are ubiquitous in the large protein sequence alignments used to predict contacts in 3D structure, as predicted by our theory. This suggests that to decouple phylogenetic effects from the interactions between sequence distal sites that control biological function, it is necessary to remove or down-weight the eigenvectors of the covariance matrix with largest eigenvalues. We confirm that truncating these eigenvectors improves contact prediction.

A Hybrid Approach for Protein Structure Determination Combining Sparse NMR with Evolutionary Coupling Sequence Data

While 3D structure determination of small (<15 kDa) proteins by solution NMR is largely automated and routine, structural analysis of larger proteins is more challenging. An emerging hybrid strategy for modeling protein structures combines sparse NMR data that can be obtained for larger proteins with sequence co-variation data, called evolutionary couplings (ECs), obtained from multiple sequence alignments of protein families. This hybrid "EC-NMR" method can be used to accurately model larger (15-60 kDa) proteins, and more rapidly determine structures of smaller (5-15 kDa) proteins using only backbone NMR data. The resulting structures have accuracies relative to reference structures comparable to those obtained with full backbone and sidechain NMR resonance assignments. The requirement that evolutionary couplings (ECs) are consistent with NMR data recorded on a specific member of a protein family, under specific conditions, potentially also allows identification of ECs that reflect alternative allosteric or excited states of the protein structure.

Co-evolution techniques are reshaping the way we do structural bioinformatics

Co-evolution techniques were originally conceived to assist in protein structure prediction by inferring pairs of residues that share spatial proximity. However, the functional relationships that can be extrapolated from co-evolution have also proven to be useful in a wide array of structural bioinformatics applications. These techniques are a powerful way to extract structural and functional information in a sequence-rich world.

Widespread Historical Contingency in Influenza Viruses

In systems biology and genomics, epistasis characterizes the impact that a substitution at a particular location in a genome can have on a substitution at another location. This phenomenon is often implicated in the evolution of drug resistance or to explain why particular "disease-causing" mutations do not have the same outcome in all individuals. Hence, uncovering these mutations and their locations in a genome is a central question in biology. However, epistasis is notoriously difficult to uncover, especially in fast-evolving organisms. Here, we present a novel statistical approach that replies on a model developed in ecology and that we adapt to analyze genetic data in fast-evolving systems such as the influenza A virus. We validate the approach using a two-pronged strategy: extensive simulations demonstrate a low-to-moderate sensitivity with excellent specificity and precision, while analyses of experimentally validated data recover known interactions, including in a eukaryotic system. We further evaluate the ability of our approach to detect correlated evolution during antigenic shifts or at the emergence of drug resistance. We show that in all cases, correlated evolution is prevalent in influenza A viruses, involving many pairs of sites linked together in chains; a hallmark of historical contingency. Strikingly, interacting sites are separated by large physical distances, which entails either long-range conformational changes or functional tradeoffs, for which we find support with the emergence of drug resistance. Our work paves a new way for the unbiased detection of epistasis in a wide range of organisms by performing whole-genome scans.

Connecting the Sequence-Space of Bacterial Signaling Proteins to Phenotypes Using Coevolutionary Landscapes

Two-component signaling (TCS) is the primary means by which bacteria sense and respond to the environment. TCS involves two partner proteins working in tandem, which interact to perform cellular functions whereas limiting interactions with non-partners (i.e., cross-talk). We construct a Potts model for TCS that can quantitatively predict how mutating amino acid identities affect the interaction between TCS partners and non-partners. The parameters of this model are inferred directly from protein sequence data. This approach drastically reduces the computational complexity of exploring the sequence-space of TCS proteins. As a stringent test, we compare its predictions to a recent comprehensive mutational study, which characterized the functionality of 20⁴ mutational variants of the PhoQ kinase in Escherichia coli We find that our best predictions accurately reproduce the amino acid combinations found in experiment, which enable functional signaling with its partner PhoP. These predictions demonstrate the evolutionary pressure to preserve the interaction between TCS partners as well as prevent unwanted cross-talk. Further, we calculate the mutational change in the binding affinity between PhoQ and PhoP, providing an estimate to the amount of destabilization needed to disrupt TCS.

New encouraging developments in contact prediction: Assessment of the CASP11 results

This article provides a report on the state-of-the-art in the prediction of intra-molecular residue-residue contacts in proteins based on the assessment of the predictions submitted to the CASP11 experiment. The assessment emphasis is placed on the accuracy in predicting long-range contacts. Twenty-nine groups participated in contact prediction in CASP11. At least eight of them used the recently developed evolutionary coupling techniques, with the top group (CONSIP2) reaching precision of 27% on target proteins that could not be modeled by homology. This result indicates a breakthrough in the development of methods based on the correlated mutation approach. Successful prediction of contacts was shown to be practically helpful in modeling three-dimensional structures; in particular target T0806 was modeled exceedingly well with accuracy not yet seen for ab initio targets of this size (>250 residues). Proteins 2016; 84(Suppl 1):131-144. © 2015 Wiley Periodicals, Inc.

KScons: a Bayesian approach for protein residue contact prediction using the knob-socket model of protein tertiary structure

MOTIVATION

By simplifying the many-bodied complexity of residue packing into patterns of simple pairwise secondary structure interactions between a single knob residue with a three-residue socket, the knob-socket construct allows a more direct incorporation of structural information into the prediction of residue contacts. By modeling the preferences between the amino acid composition of a socket and knob, we undertake an investigation of the knob-socket construct's ability to improve the prediction of residue contacts. The statistical model considers three priors and two posterior estimations to better understand how the input data affects predictions. This produces six implementations of KScons that are tested on three sets: PSICOV, CASP10 and CASP11. We compare against the current leading contact prediction methods.

RESULTS

The results demonstrate the usefulness as well as the limits of knob-socket based structural modeling of protein contacts. The construct is able to extract good predictions from known structural homologs, while its performance degrades when no homologs exist. Among our six implementations, KScons MST-MP (which uses the multiple structure alignment prior and marginal posterior incorporating structural homolog information) performs the best in all three prediction sets. An analysis of recall and precision finds that KScons MST-MP improves accuracy not only by improving identification of true positives, but also by decreasing the number of false positives. Over the CASP10 and CASP11 sets, KScons MST-MP performs better than the leading methods using only evolutionary coupling data, but not quite as well as the supervised learning methods of MetaPSICOV and CoinDCA-NN that incorporate a large set of structural features.

CONTACT

qiwei.li@rice.eduSupplementary information: Supplementary data are available at Bioinformatics online.

Correlated positions in protein evolution and engineering

Statistical analysis of a protein multiple sequence alignment can reveal groups of positions that undergo interdependent mutations throughout evolution. At these so-called correlated positions, only certain combinations of amino acids appear to be viable for maintaining proper folding, stability, catalytic activity or specificity. Therefore, it is often speculated that they could be interesting guides for semi-rational protein engineering purposes. Because they are a fingerprint from protein evolution, their analysis may provide valuable insight into a protein's structure or function and furthermore, they may also be suitable target positions for mutagenesis. Unfortunately, little is currently known about the properties of these correlation networks and how they should be used in practice. This review summarises the recent findings, opportunities and pitfalls of the concept.