Model-free inference for characterizing protein mutations through a coevolutionary lens

Co-evolution has an important function in the evolution of species and it is clearly manifested in certain scenarios such as host–parasite and predator–prey interactions, symbiosis and mutualism. The extrapolation of the concepts and methodologies developed for the study of species co-evolution at the molecular level has prompted the development of a variety of computational methods able to predict protein interactions through the characteristics of co-evolution. Particularly successful have been those methods that predict interactions at the genomic level based on the detection of pairs of protein families with similar evolutionary histories (similarity of phylogenetic trees: mirrortree). Future advances in this field will require a better understanding of the molecular basis of the co-evolution of protein families. Thus, it will be important to decipher the molecular mechanisms underlying the similarity observed in phylogenetic trees of interacting proteins, distinguishing direct specific molecular interactions from other general functional constraints. In particular, it will be important to separate the effects of physical interactions within protein complexes (‘co-adaptation') from other forces that, in a less specific way, can also create general patterns of co-evolution.

Enhancing Evolutionary Couplings with Deep Convolutional Neural Networks.

Motivation: Recent breakthroughs in protein residue–residue contact prediction have made reliable de novo prediction of protein structures possible. The key was to apply statistical methods that can distinguish direct couplings between pairs of columns in a multiple sequence alignment from merely correlated pairs, i.e. to separate direct from indirect effects. Two classes of such methods exist, either relying on regularized inversion of the covariance matrix or on pseudo-likelihood maximization (PLM). Although PLM-based methods offer clearly higher precision, available tools are not sufficiently optimized and are written in interpreted languages that introduce additional overheads. This impedes the runtime and large-scale contact prediction for larger protein families, multi-domain proteins and protein–protein interactions.Results: Here we introduce CCMpred, our performance-optimized PLM implementation in C and CUDA C. Using graphics cards in the price range of current six-core processors, CCMpred can predict contacts for typical alignments 35–113 times faster and with the same precision as the most accurate published methods. For users without a CUDA-capable graphics card, CCMpred can also run in a CPU mode that is still 4–14 times faster. Thanks to our speed-ups (http://dictionary.cambridge.org/dictionary/british/speed-up) contacts for typical protein families can be predicted in 15–60 s on a consumer-grade GPU and 1–6 min on a six-core CPU.Availability and implementation: CCMpred is free and open-source software under the GNU Affero General Public License v3 (or later) available at https://bitbucket.org/soedinglab/ccmpredContact: johannes.soeding@mpibpc.mpg.deSupplementary information: Supplementary data are available at Bioinformatics online.

ComplexContact (http://raptorx2.uchicago.edu/ComplexContact/) is a web server for sequence-based interfacial residue-residue contact prediction of a putative protein complex. Interfacial residue-residue contacts are critical for understanding how proteins form complex and interact at residue level. When receiving a pair of protein sequences, ComplexContact first searches for their sequence homologs and builds two paired multiple sequence alignments (MSA), then it applies co-evolution analysis and a CASP-winning deep learning (DL) method to predict interfacial contacts from paired MSAs and visualizes the prediction as an image. The DL method was originally developed for intra-protein contact prediction and performed the best in CASP12. Our large-scale experimental test further shows that ComplexContact greatly outperforms pure co-evolution methods for inter-protein contact prediction, regardless of the species.

BackgroundMuch progress has been made in understanding the 3D structure of proteins using methods such as NMR and X-ray crystallography. The resulting 3D structures are extremely informative, but do not always reveal which sites and residues within the structure are of special importance. Recently, there are indications that multiple-residue, sub-domain structural relationships within the larger 3D consensus structure of a protein can be inferred from the analysis of the multiple sequence alignment data of a protein family. These intra-dependent clusters of associated sites are used to indicate hierarchical inter-residue relationships within the 3D structure. To reveal the patterns of associations among individual amino acids or sub-domain components within the structure, we apply a k-modes attribute (aligned site) clustering algorithm to the ubiquitin and transthyretin families in order to discover associations among groups of sites within the multiple sequence alignment. We then observe what these associations imply within the 3D structure of these two protein families.ResultsThe k-modes site clustering algorithm we developed maximizes the intra-group interdependencies based on a normalized mutual information measure. The clusters formed correspond to sub-structural components or binding and interface locations. Applying this data-directed method to the ubiquitin and transthyretin protein family multiple sequence alignments as a test bed, we located numerous interesting associations of interdependent sites. These clusters were then arranged into cluster tree diagrams which revealed four structural sub-domains within the single domain structure of ubiquitin and a single large sub-domain within transthyretin associated with the interface among transthyretin monomers. In addition, several clusters of mutually interdependent sites were discovered for each protein family, each of which appear to play an important role in the molecular structure and/or function.ConclusionsOur results demonstrate that the method we present here using a k-modes site clustering algorithm based on interdependency evaluation among sites obtained from a sequence alignment of homologous proteins can provide significant insights into the complex, hierarchical inter-residue structural relationships within the 3D structure of a protein family.

Protein–protein interactions (PPIs) are fundamental to many biological processes. The coevolution-based prediction of interacting residues has made great strides in protein complexes that are known to interact. A multiple sequence alignment (MSA) is the basis of coevolution analysis. MSAs have recently made significant progress in the protein monomer sequence analysis. However, no standard or efficient pipelines are available for the sensitive protein complex MSA (cpxMSA) collection. How to generate cpxMSA is one of the most challenging problems of sequence coevolution analysis. Although several methods have been developed to address this problem, no standalone program exists. Furthermore, the number of built-in properties is limited; hence, it is often difficult for users to analyze sequence coevolution according to their desired cpxMSA. In this article, we developed a novel cpxMSA approach (cpxDeepMSA. We used different protein monomer databases and incorporated the three strategies (genomic distance, phylogeny information, and STRING interaction network) used to join the monomer MSA results of protein complexes, which can prevent using a single method fail to the joint two-monomer MSA causing the cpxMSA construction failure. We anticipate that the cpxDeepMSA algorithm will become a useful high-throughput tool in protein complex structure predictions, inter-protein residue-residue contacts, and the biological sequence coevolution analysis.

The latent features extracted from the multiple sequence alignments (MSAs) of homologous protein families are useful for identifying residue-residue contacts, predicting mutation effects, shaping protein evolution, etc. Over the past three decades, a growing body of supervised and unsupervised machine learning methods have been applied to this field, yielding fruitful results. Here, we propose a novel self-supervised model, called encoder-transformation layer-decoder (ETLD) architecture, capable of capturing protein sequence latent features directly from MSAs. Compared to the typical autoencoder model, ETLD introduces a transformation layer with the ability to learn inter-site couplings, which can be used to parse out the two-dimensional residue-residue contacts map after a simple mathematical derivation or an additional supervised neural network. ETLD retains the process of encoding and decoding sequences, and the predicted probabilities of amino acids at each site can be further used to construct the mutation landscapes for mutation effects prediction, outperforming advanced models such as GEMME, DeepSequence and EVmutation in general. Overall, ETLD is a highly interpretable unsupervised model with great potential for improvement and can be further combined with supervised methods for more extensive and accurate predictions.

Blocks are ungapped multiple sequence alignments representing conserved protein regions, and the Blocks Database consists of blocks from documented protein families. World Wide Web (http://www. blocks.fhcrc.org) and Email (blocks@blocks.fhcrc.org) servers provide tools for homology searching and for analyzing protein family relationships. New enhancements include a multiple alignment processor that extends the use of these tools to imported multiple alignments of families not present in the database and a PCR primer designer that implements a new strategy for gene isolation.

The multiple sequence alignment (MSA) is the entry point of many RNA structure modeling tasks, such as prediction of RNA secondary structure (rSS) and contacts. However, there are few automated programs for generating high quality MSAs of target RNA molecules. We have developed rMSA, a hierarchical pipeline for sensitive search and accurate alignment of RNA homologs for a target RNA. On a diverse set of 365 non-redundant RNA structures, rMSA significantly outperforms an existing MSA generation method (RNAcmap) by approximately 20% and 5% higher F1-scores for rSS and long-range contact prediction, respectively. rMSA is available at https://zhanggroup.org/rMSA/ and https://github.com/pylelab/rMSA.

The standard approach for constructing a phylogenetic tree from a set of sequences consists of two key stages. First, a multiple sequence alignment (MSA) of the sequences is computed. The aligned data are then used to reconstruct the phylogenetic tree. The accuracy of the resulting tree heavily relies on the quality of the MSA. The quality of the popularly used progressive sequence alignment depends on a guide tree, which determines the order of aligning sequences. Most MSA methods use pairwise comparisons to generate a distance matrix and reconstruct the guide tree. However, when dealing with highly diverged sequences, constructing a good guide tree is challenging. In this work, we propose an alternative approach using three-way dynamic programming alignment to generate the distance matrix and the guide tree. This three-way alignment incorporates information from additional sequences to compute evolutionary distances more accurately. Using simulated datasets on two symmetric and asymmetric trees, we compared MAFFT with its default guide tree with MAFFT with a guide tree produced using the three-way alignment. We found that (1) the three-way alignment can reconstruct better guide trees than those from the most accurate options of MAFFT, and (2) the better guide tree, on average, leads to more accurate phylogenetic reconstruction. However, the improvement over the L-INS-i option of MAFFT is small, attesting to the excellence of the alignment quality of MAFFT. Surprisingly, the two criteria for choosing the best MSA (phylogenetic accuracy and sum-of-pair score) conflict with each other.

Ranking Docking Models by Coevolution Analysis

To understand the functional elements of a protein structure biologists use domain specific 3D viewers (PDB) that are written to process the coordinates of atoms that represent the solved protein structure using X-Ray crystallography or NMR. The PDB viewers have been written to capture specific or common features of interest to the researcher. With the explosion of protein sequence data comparative studies and statistical analysis of data can indicate regions of interest in 3D models. The ability to integrate statistical data into existing PDB viewers is difficult because the software is typically written to accomplish very specific functional goals and does not support exporting to a standard 3D format. In this paper, the PDB data is shown as X3D PDB ribbon models that are augmented with statistically significant data and compared to an Information-Rich Virtual Environment represented as a Protein CorreLogo X3D model.A protein family (Pfam) represents multiple alignments of protein sequences where protein domains and the tertiary structures have evolutionary conserved regions representing protein function. Various information properties of the protein family, the tertiary structure from a sequence's PDB structure and ligand binding pockets are combined to create a 3D Protein CorreLogo model. The multiple sequence alignment from the protein family is used to detect co-evolving amino acid pairs using mutual information. Co-evolving pairs are indicated as a column with color coding to represent the physio-chemical properties of each co-evolving amino acid combination. Additional visualizations along each axis include the 2D sequence logo, the degree of insert regions in the protein family and the surface accessibility of each amino acid for the referenced PDB sequence. The Protein CorreLogo model is based on X3D (VRML) facilitating immersive viewing of complex data relationships and detected co-evolving pairs.Two protein families are presented in the results section that compare the Protein CorreLogo model with a representative X3D RDB ribbon model showing the structural significance of predicted co-evolving amino acid pairs using mutual information. One example protein family, with proteins that bind cyclic nucleotides (PF00027.18), is given where the co-evolving pairs are potential markers for ligand binding pocket regions. Another example protein family, with SH3 domains that are involved in signal transduction related to cytoskeletal organization (PF00018.16), shows significant mutual information occurring between two pairs of amino acids that are in contact in the intertwined dimer structure but are on opposite ends of the tertiary structure.Protein CorreLogo X3D models and X3D PDB ribbon models can be found at http://www.proteinx3d.com

The profile hidden Markov model (PHMM) is widely used to assign the protein sequences to their respective families. A major limitation of a PHMM is the assumption that given states the observations (amino acids) are independent. To overcome this limitation, the dependency between amino acids in a multiple sequence alignment (MSA) which is the representative of a PHMM can be appended to the PHMM. Due to the fact that with a MSA, the sequences of amino acids are biologically related, the one-by-one dependency between two amino acids can be considered. In other words, based on the MSA, the dependency between an amino acid and its corresponding amino acid located above can be combined with the PHMM. For this purpose, the new emission probability matrix which considers the one-by-one dependencies between amino acids is constructed. The parameters of a PHMM are of two types; transition and emission probabilities which are usually estimated using an EM algorithm called the Baum-Welch algorithm. We have generalized the Baum-Welch algorithm using similarity emission matrix constructed by integrating the new emission probability matrix with the common emission probability matrix. Then, the performance of similarity emission is discussed by applying it to the top twenty protein families in the Pfam database. We show that using the similarity emission in the Baum-Welch algorithm significantly outperforms the common Baum-Welch algorithm in the task of assigning protein sequences to protein families.

A fundamental problem in molecular biology is the prediction of the three-dimensional structure of a protein from its amino acid sequence. However, molecular modeling to find the structure is at present intractable and is likely to remain so for some time, hence intermediate steps such as predicting which residues pairs are in contact have been developed. Predicted contact pairs have been used for fold prediction, as an initial condition or constraint for molecular modeling, and as a filter to rank multiple models arising from homology modeling. As contact prediction has advanced it is becoming more common for 3D structure predictors to integrate contact prediction into structure building, as this often gives information that is orthogonal to that produced by other methods. This chapter shows how evolutionary information contained in protein sequences and multiple sequence alignments can be used to predict protein structure, and the state-of-the-art predictors and their methodologies are reviewed.

Contact-map of a protein sequence dictates the global topology of structural fold. Accurate prediction of the contact-map is thus essential to protein 3D structure prediction, which is particularly useful for the protein sequences that do not have close homology templates in the Protein Data Bank. We developed a new method, ResPRE, to predict residue-level protein contacts using inverse covariance matrix (or precision matrix) of multiple sequence alignments (MSAs) through deep residual convolutional neural network training. The approach was tested on a set of 158 non-homologous proteins collected from the CASP experiments and achieved an average accuracy of 50.6% in the top-L long-range contact prediction with L being the sequence length, which is 11.7% higher than the best of other state-of-the-art approaches ranging from coevolution coupling analysis to deep neural network training. Detailed data analyses show that the major advantage of ResPRE lies at the utilization of precision matrix that helps rule out transitional noises of contact-maps compared with the previously used covariance matrix. Meanwhile, the residual network with parallel shortcut layer connections increases the learning ability of deep neural network training. It was also found that appropriate collection of MSAs can further improve the accuracy of final contact-map predictions. The standalone package and online server of ResPRE are made freely available, which should bring important impact on protein structure and function modeling studies in particular for the distant- and non-homology protein targets. https://zhanglab.ccmb.med.umich.edu/ResPRE and https://github.com/leeyang/ResPRE. Supplementary data are available at Bioinformatics online.