Generative machine learning models are increasingly being used to design novel proteins. However, their major limitation is the inability to account for protein flexibility, a property crucial for protein function. Learning to engineer flexibility is difficult because the relevant data is scarce, heterogeneous, and costly to obtain using computational and experimental methods. Our contributions are three-fold. First, we perform a comprehensive comparison of methods for evaluating protein flexibility and identify relevant data for learning. Second, we overcome the data scarcity issue by leveraging a pre-trained protein language model. We design and train flexibility predictors utilizing either only sequential or both sequential and structural information on the input. Third, we introduce a method for fine-tuning a protein inverse folding model to make it steerable toward desired flexibility at specified regions. We demonstrate that our method Flexpert enables guidance of inverse folding models toward increased flexibility. This opens up a transformative possibility of engineering protein flexibility.

Discovering mutations enhancing protein-protein interactions (PPIs) is critical for advancing biomedical research and developing improved therapeutics. While machine learning approaches have substantially advanced the field, they often struggle to generalize beyond training data in practical scenarios. The contributions of this work are three-fold. First, we construct PPIRef, the largest and non-redundant dataset of 3D protein-protein interactions, enabling effective large-scale learning. Second, we leverage the PPIRef dataset to pre-train PPIformer, a new SE(3)-equivariant model generalizing across diverse protein-binder variants. We fine-tune PPIformer to predict effects of mutations on protein-protein interactions via a thermodynamically motivated adjustment of the pre-training loss function. Finally, we demonstrate the enhanced generalization of our new PPIformer approach by outperforming other state-of-the-art methods on new, non-leaking splits of standard labeled PPI mutational data and independent case studies optimizing a human antibody against SARS-CoV-2 and increasing the thrombolytic activity of staphylokinase.

Thermostable proteins find their use in numerous biomedical and biotechnological applications. However, the computational design of stable proteins often results in single-point mutations with a limited effect on protein stability. However, the construction of stable multiple-point mutants can prove difficult due to the possibility of antagonistic effects between individual mutations. FireProt protocol enables the automated computational design of highly stable multiple-point mutants. FireProt 2.0 builds on top of the previously published FireProt web, retaining the original functionality and expanding it with several new stabilization strategies. FireProt 2.0 integrates the AlphaFold database and the homology modeling for structure prediction, enabling calculations starting from a sequence. Multiple-point designs are constructed using the Bron-Kerbosch algorithm minimizing the antagonistic effect between the individual mutations. Users can newly limit the FireProt calculation to a set of user-defined mutations, run a saturation mutagenesis of the whole protein or select rigidifying mutations based on B-factors. Evolution-based back-to-consensus strategy is complemented by ancestral sequence reconstruction. FireProt 2.0 is significantly faster and a reworked graphical user interface broadens the tool's availability even to users with older hardware. FireProt 2.0 is freely available at http://loschmidt.chemi.muni.cz/fireprotweb.

PredictONCO 1.0 is a unique web server that analyzes effects of mutations on proteins frequently altered in various cancer types. The server can assess the impact of mutations on the protein sequential and structural properties and apply a virtual screening to identify potential inhibitors that could be used as a highly individualized therapeutic approach, possibly based on the drug repurposing. PredictONCO integrates predictive algorithms and state-of-the-art computational tools combined with information from established databases. The user interface was carefully designed for the target specialists in precision oncology, molecular pathology, clinical genetics and clinical sciences. The tool summarizes the effect of the mutation on protein stability and function and currently covers 44 common oncological targets. The binding affinities of Food and Drug Administration/ European Medicines Agency -approved drugs with the wild-type and mutant proteins are calculated to facilitate treatment decisions. The reliability of predictions was confirmed against 108 clinically validated mutations. The server provides a fast and compact output, ideal for the often time-sensitive decision-making process in oncology. Three use cases of missense mutations, (i) K22A in cyclin-dependent kinase 4 identified in melanoma, (ii) E1197K mutation in anaplastic lymphoma kinase 4 identified in lung carcinoma and (iii) V765A mutation in epidermal growth factor receptor in a patient with congenital mismatch repair deficiency highlight how the tool can increase levels of confidence regarding the pathogenicity of the variants and identify the most effective inhibitors. The server is available at https://loschmidt.chemi.muni.cz/predictonco.

SUMMARY: Protein design requires information about how mutations affect protein stability. Many web-based predictors are available for this purpose, yet comparing them or using them en masse is difficult. Here, we present BenchStab, a console tool/Python package for easy and quick execution of 19 predictors and result collection on a list of mutants. Moreover, the tool is easily extensible with additional predictors. We created an independent dataset derived from the FireProtDB and evaluated 24 different prediction methods. AVAILABILITY AND IMPLEMENTATION: BenchStab is an open-source Python package available at https://github.com/loschmidt/BenchStab with a detailed README and example usage at https://loschmidt.chemi.muni.cz/benchstab. The BenchStab dataset is available on Zenodo: https://zenodo.org/records/10637728.

SUMMARY: Access pathways in enzymes are crucial for the passage of substrates and products of catalysed reactions. The process can be studied by computational means with variable degrees of precision. Our in-house approximative method CaverDock provides a fast and easy way to set up and run ligand binding and unbinding calculations through protein tunnels and channels. Here we introduce pyCaverDock, a Python3 API designed to improve user experience with the tool and further facilitate the ligand transport analyses. The API enables users to simplify the steps needed to use CaverDock, from automatizing setup processes to designing screening pipelines. AVAILABILITY AND IMPLEMENTATION: pyCaverDock API is implemented in Python 3 and is freely available with detailed documentation and practical examples at https://loschmidt.chemi.muni.cz/caverdock/.

There is a great interest in increasing proteins' stability to widen their usability in numerous biomedical and biotechnological applications. However, native proteins cannot usually withstand the harsh industrial environment, since they are evolved to function under mild conditions. Ancestral sequence reconstruction is a well-established method for deducing the evolutionary history of genes. Besides its applicability to discover the most probable evolutionary ancestors of the modern proteins, ancestral sequence reconstruction has proven to be a useful approach for the design of highly stable proteins. Recently, several computational tools were developed, which make the ancestral reconstruction algorithms accessible to the community, while leaving the most crucial steps of the preparation of the input data on users' side. FireProtASR aims to overcome this obstacle by constructing a fully automated workflow, allowing even the unexperienced users to obtain ancestral sequences based on a sequence query as the only input. FireProtASR is complemented with an interactive, easy-to-use web interface and is freely available at https://loschmidt.chemi.muni.cz/fireprotasr/.

MOTIVATION: Poor protein solubility hinders the production of many therapeutic and industrially useful proteins. Experimental efforts to increase solubility are plagued by low success rates and often reduce biological activity. Computational prediction of protein expressibility and solubility in Escherichia coli using only sequence information could reduce the cost of experimental studies by enabling prioritization of highly soluble proteins. RESULTS: A new tool for sequence-based prediction of soluble protein expression in E.coli, SoluProt, was created using the gradient boosting machine technique with the TargetTrack database as a training set. When evaluated against a balanced independent test set derived from the NESG database, SoluProt's accuracy of 58.5% and AUC of 0.62 exceeded those of a suite of alternative solubility prediction tools. There is also evidence that it could significantly increase the success rate of experimental protein studies. SoluProt is freely available as a standalone program and a user-friendly webserver at https://loschmidt.chemi.muni.cz/soluprot/. AVAILABILITY AND IMPLEMENTATION: https://loschmidt.chemi.muni.cz/soluprot/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Abstract Computation of trajectories for ligand binding and unbinding via protein tunnels and channels is important for predicting possible protein–ligand interactions. These highly complex processes can be simulated by several software tools, which provide biochemists with valuable information for drug design or protein engineering applications. This paper focuses on aiding this exploration process by introducing the DockVis visual analysis tool. DockVis operates with the multivariate output data from one of the latest available tools for the prediction of ligand transport, CaverDock. DockVis provides the users with several linked views, combining the 2D abstracted depictions of ligands and their surroundings and properties with the 3D view. In this way, we enable the users to perceive the spatial configurations of ligand passing through the protein tunnel. The users are initially visually directed to the most relevant parts of ligand trajectories, which can be then explored in higher detail by the follow‐up analyses. DockVis was designed in tight collaboration with protein engineers developing the CaverDock tool. However, the concept of DockVis can be extended to any other tool predicting ligand pathways by the molecular docking. DockVis will be made available to the wide user community as part of the Caver Analyst 3.0 software package ( www.caver.cz ).

Here we present a novel method for the analysis of transport processes in proteins and its implementation called CaverDock. Our method is based on a modified molecular docking algorithm. It iteratively places the ligand along the access tunnel in such a way that the ligand movement is contiguous and the energy is minimized. The result of CaverDock calculation is a ligand trajectory and an energy profile of transport process. CaverDock uses the modified docking program Autodock Vina for molecular docking and implements a parallel heuristic algorithm for searching the space of possible trajectories. Our method lies in between the geometrical approaches and molecular dynamics simulations. Contrary to the geometrical methods, it provides an evaluation of chemical forces. However, it is far less computationally demanding and easier to set up compared to molecular dynamics simulations. CaverDock will find a broad use in the fields of computational enzymology, drug design, and protein engineering. The software is available free of charge to the academic users at https://loschmidt.chemi.muni.cz/caverdock/.

MOTIVATION: Protein tunnels and channels are key transport pathways that allow ligands to pass between proteins' external and internal environments. These functionally important structural features warrant detailed attention. It is difficult to study the ligand binding and unbinding processes experimentally, while molecular dynamics simulations can be time-consuming and computationally demanding. RESULTS: CaverDock is a new software tool for analysing the ligand passage through the biomolecules. The method uses the optimized docking algorithm of AutoDock Vina for ligand placement docking and implements a parallel heuristic algorithm to search the space of possible trajectories. The duration of the simulations takes from minutes to a few hours. Here we describe the implementation of the method and demonstrate CaverDock's usability by: (i) comparison of the results with other available tools, (ii) determination of the robustness with large ensembles of ligands and (iii) the analysis and comparison of the ligand trajectories in engineered tunnels. Thorough testing confirms that CaverDock is applicable for the fast analysis of ligand binding and unbinding in fundamental enzymology and protein engineering. AVAILABILITY AND IMPLEMENTATION: User guide and binaries for Ubuntu are freely available for non-commercial use at https://loschmidt.chemi.muni.cz/caverdock/. The web implementation is available at https://loschmidt.chemi.muni.cz/caverweb/. The source code is available upon request. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Abstract Analyzing molecular dynamics (MD) simulations is a key aspect to understand protein dynamics and function. With increasing computational power, it is now possible to generate very long and complex simulations, which are cumbersome to explore using traditional 3D animations of protein movements. Guided by requirements derived from multiple focus groups with protein engineering experts, we designed and developed a novel interactive visual analysis approach for long and crowded MD simulations. In this approach, we link a dynamic 3D focus+context visualization with a 2D chart of time series data to guide the detection and navigation towards important spatio‐temporal events. The 3D visualization renders elements of interest in more detail and increases the temporal resolution dependent on the time series data or the spatial region of interest. In case studies with different MD simulation data sets and research questions, we found that the proposed visual analysis approach facilitates exploratory analysis to generate, confirm, or reject hypotheses about causalities. Finally, we derived design guidelines for interactive visual analysis of complex MD simulation data.

Motivation: Studying the transport paths of ligands, solvents, or ions in transmembrane proteins and proteins with buried binding sites is fundamental to the understanding of their biological function. A detailed analysis of the structural features influencing the transport paths is also important for engineering proteins for biomedical and biotechnological applications. Results: CAVER Analyst 2.0 is a software tool for quantitative analysis and real-time visualization of tunnels and channels in static and dynamic structures. This version provides the users with many new functions, including advanced techniques for intuitive visual inspection of the spatiotemporal behavior of tunnels and channels. Novel integrated algorithms allow an efficient analysis and data reduction in large protein structures and molecular dynamic simulations. Availability and implementation: CAVER Analyst 2.0 is a multi-platform standalone Java-based application. Binaries and documentation are freely available at www.caver.cz. Supplementary information: Supplementary data are available at Bioinformatics online.

An important message taken from human genome sequencing projects is that the human population exhibits approximately 99.9% genetic similarity. Variations in the remaining parts of the genome determine our identity, trace our history and reveal our heritage. The precise delineation of phenotypically causal variants plays a key role in providing accurate personalized diagnosis, prognosis, and treatment of inherited diseases. Several computational methods for achieving such delineation have been reported recently. However, their ability to pinpoint potentially deleterious variants is limited by the fact that their mechanisms of prediction do not account for the existence of different categories of variants. Consequently, their output is biased towards the variant categories that are most strongly represented in the variant databases. Moreover, most such methods provide numeric scores but not binary predictions of the deleteriousness of variants or confidence scores that would be more easily understood by users. We have constructed three datasets covering different types of disease-related variants, which were divided across five categories: (i) regulatory, (ii) splicing, (iii) missense, (iv) synonymous, and (v) nonsense variants. These datasets were used to develop category-optimal decision thresholds and to evaluate six tools for variant prioritization: CADD, DANN, FATHMM, FitCons, FunSeq2 and GWAVA. This evaluation revealed some important advantages of the category-based approach. The results obtained with the five best-performing tools were then combined into a consensus score. Additional comparative analyses showed that in the case of missense variations, protein-based predictors perform better than DNA sequence-based predictors. A user-friendly web interface was developed that provides easy access to the five tools' predictions, and their consensus scores, in a user-understandable format tailored to the specific features of different categories of variations. To enable comprehensive evaluation of variants, the predictions are complemented with annotations from eight databases. The web server is freely available to the community at http://loschmidt.chemi.muni.cz/predictsnp2.

The biological function of a macromolecule often requires that a small molecule or ion is transported through its structure. The transport pathway often leads through void spaces in the structure. The properties of transport pathways change significantly in time; therefore, the analysis of a trajectory from molecular dynamics rather than of a single static structure is needed for understanding the function of pathways. The identification and analysis of transport pathways are challenging because of the high complexity and diversity of macromolecular shapes, the thermal motion of their atoms, and the large amount of conformations needed to properly describe conformational space of protein structure. In this paper, we describe the principles of the CAVER 3.0 algorithms for the identification and analysis of properties of transport pathways both in static and dynamic structures. Moreover, we introduce the improved clustering solution for finding tunnels in macromolecules, which is included in the latest CAVER 3.02 version. Voronoi diagrams are used to identify potential pathways in each snapshot of a molecular dynamics trajectory and clustering is then used to find the correspondence between tunnels from different snapshots. Furthermore, the geometrical properties of pathways and their evolution in time are computed and visualized.

There is great interest in increasing proteins' stability to enhance their utility as biocatalysts, therapeutics, diagnostics and nanomaterials. Directed evolution is a powerful, but experimentally strenuous approach. Computational methods offer attractive alternatives. However, due to the limited reliability of predictions and potentially antagonistic effects of substitutions, only single-point mutations are usually predicted in silico, experimentally verified and then recombined in multiple-point mutants. Thus, substantial screening is still required. Here we present FireProt, a robust computational strategy for predicting highly stable multiple-point mutants that combines energy- and evolution-based approaches with smart filtering to identify additive stabilizing mutations. FireProt's reliability and applicability was demonstrated by validating its predictions against 656 mutations from the ProTherm database. We demonstrate that thermostability of the model enzymes haloalkane dehalogenase DhaA and γ-hexachlorocyclohexane dehydrochlorinase LinA can be substantially increased (ΔTm = 24°C and 21°C) by constructing and characterizing only a handful of multiple-point mutants. FireProt can be applied to any protein for which a tertiary structure and homologous sequences are available, and will facilitate the rapid development of robust proteins for biomedical and biotechnological applications.

Abstract Summary: The transport of ligands, ions or solvent molecules into proteins with buried binding sites or through the membrane is enabled by protein tunnels and channels. CAVER Analyst is a software tool for calculation, analysis and real-time visualization of access tunnels and channels in static and dynamic protein structures. It provides an intuitive graphic user interface for setting up the calculation and interactive exploration of identified tunnels/channels and their characteristics. Availability and Implementation: CAVER Analyst is a multi-platform software written in JAVA. Binaries and documentation are freely available for non-commercial use at http://www.caver.cz . Contact: [email protected] or [email protected] Supplementary information: Supplementary data are available at Bioinformatics online.

Single nucleotide variants represent a prevalent form of genetic variation. Mutations in the coding regions are frequently associated with the development of various genetic diseases. Computational tools for the prediction of the effects of mutations on protein function are very important for analysis of single nucleotide variants and their prioritization for experimental characterization. Many computational tools are already widely employed for this purpose. Unfortunately, their comparison and further improvement is hindered by large overlaps between the training datasets and benchmark datasets, which lead to biased and overly optimistic reported performances. In this study, we have constructed three independent datasets by removing all duplicities, inconsistencies and mutations previously used in the training of evaluated tools. The benchmark dataset containing over 43,000 mutations was employed for the unbiased evaluation of eight established prediction tools: MAPP, nsSNPAnalyzer, PANTHER, PhD-SNP, PolyPhen-1, PolyPhen-2, SIFT and SNAP. The six best performing tools were combined into a consensus classifier PredictSNP, resulting into significantly improved prediction performance, and at the same time returned results for all mutations, confirming that consensus prediction represents an accurate and robust alternative to the predictions delivered by individual tools. A user-friendly web interface enables easy access to all eight prediction tools, the consensus classifier PredictSNP and annotations from the Protein Mutant Database and the UniProt database. The web server and the datasets are freely available to the academic community at http://loschmidt.chemi.muni.cz/predictsnp.

Tunnels and channels facilitate the transport of small molecules, ions and water solvent in a large variety of proteins. Characteristics of individual transport pathways, including their geometry, physico-chemical properties and dynamics are instrumental for understanding of structure-function relationships of these proteins, for the design of new inhibitors and construction of improved biocatalysts. CAVER is a software tool widely used for the identification and characterization of transport pathways in static macromolecular structures. Herein we present a new version of CAVER enabling automatic analysis of tunnels and channels in large ensembles of protein conformations. CAVER 3.0 implements new algorithms for the calculation and clustering of pathways. A trajectory from a molecular dynamics simulation serves as the typical input, while detailed characteristics and summary statistics of the time evolution of individual pathways are provided in the outputs. To illustrate the capabilities of CAVER 3.0, the tool was applied for the analysis of molecular dynamics simulation of the microbial enzyme haloalkane dehalogenase DhaA. CAVER 3.0 safely identified and reliably estimated the importance of all previously published DhaA tunnels, including the tunnels closed in DhaA crystal structures. Obtained results clearly demonstrate that analysis of molecular dynamics simulation is essential for the estimation of pathway characteristics and elucidation of the structural basis of the tunnel gating. CAVER 3.0 paves the way for the study of important biochemical phenomena in the area of molecular transport, molecular recognition and enzymatic catalysis. The software is freely available as a multiplatform command-line application at http://www.caver.cz.

BACKGROUND: The main aim of this study was to develop and implement an algorithm for the rapid, accurate and automated identification of paths leading from buried protein clefts, pockets and cavities in dynamic and static protein structures to the outside solvent. RESULTS: The algorithm to perform a skeleton search was based on a reciprocal distance function grid that was developed and implemented for the CAVER program. The program identifies and visualizes routes from the interior of the protein to the bulk solvent. CAVER was primarily developed for proteins, but the algorithm is sufficiently robust to allow the analysis of any molecular system, including nucleic acids or inorganic material. Calculations can be performed using discrete structures from crystallographic analysis and NMR experiments as well as with trajectories from molecular dynamics simulations. The fully functional program is available as a stand-alone version and as plug-in for the molecular modeling program PyMol. Additionally, selected functions are accessible in an online version. CONCLUSION: The algorithm developed automatically finds the path from a starting point located within the interior of a protein. The algorithm is sufficiently rapid and robust to enable routine analysis of molecular dynamics trajectories containing thousands of snapshots. The algorithm is based on reciprocal metrics and provides an easy method to find a centerline, i.e. the spine, of complicated objects such as a protein tunnel. It can also be applied to many other molecules. CAVER is freely available from the web site http://loschmidt.chemi.muni.cz/caver/.

MOTIVATION: One of the objectives of protein engineering is to propose and construct modified proteins with improved activity for the substrate of interest. Systematic computational investigation of many protein variants requires the preparation and handling of a large number of data files. The type of the data generated during the modelling of protein variants and the estimation of their activities offers the possibility of process automatization. RESULTS: The graphical program TRITON has been developed for modelling protein mutants and assessment of their activities. Protein mutants are modelled from the wild type structure by homology modelling using the external program MODELLER. Chemical reactions taking place in the mutants active site are modelled using the semi-empirical quantum mechanic program MOPAC. Semi-quantitative predictions of mutants activities can be achieved by evaluating the changes in energies of the system and partial atomic charges of active site residues during the reaction. The program TRITON offers graphical tools for the preparation of the input data files, for calculation and for the analysis of the generated output data. AVAILABILITY: The program TRITON can run under operating systems IRIX, Linux and NetBSD. The software is available at http://www.chemi.muni.cz/lbsd/triton.ht ml.