The EVcouplings Python framework for coevolutionary sequence analysis.

SUMMARY: Coevolutionary sequence analysis has become a commonly used technique for de novo prediction of the structure and function of proteins, RNA, and protein complexes. We present the EVcouplings framework, a fully integrated open-source application and Python package for coevolutionary analysis. The framework enables generation of sequence alignments, calculation and evaluation of evolutionary couplings (ECs), and de novo prediction of structure and mutation effects. The combination of an easy to use, flexible command line interface and an underlying modular Python package makes the full power of coevolutionary analyses available to entry-level and advanced users.
AVAILABILITY AND IMPLEMENTATION: https://github.com/debbiemarkslab/evcouplings.

1 Introduction

Coevolutionary sequence analysis presents a promising new approach to the long-standing problem of de novo prediction of the 3D structure of proteins and RNAs. In this approach, pairwise graphical models are used to identify evolutionary couplings (ECs) between sites, which frequently correspond to physical contacts in the molecule’s 3D structure. ECs have been used to successfully predict the residue contacts (Balakrishnan ; Ekeberg ; Marks ; Morcos ) and full 3D structure of proteins (Hopf ; Marks ; Ovchinnikov ), RNAs (Weinreb ), complexes (Hopf ; Ovchinnikov ; Weigt ), as well as effects of mutations (Figliuzzi ; Hopf ). However, these applications require integrating multiple tools, data sources and extensive data processing. Available software in this field provides high-performance reimplementations of EC inference tools (Kaján ; Seemayer ; Weinreb ), integration of multiple signals to improve prediction accuracy (Jones ; Skwark ), and a library targeted at format conversion between the different approaches (Simkovic ). To make these methods accessible to a general biological audience, we present a flexible, open source application and Python package for end-to-end evolutionary coupling analysis. EVcouplings, making use of external tools, covers all necessary functionality, including alignment generation, EC calculation, de novo structure and mutation effect prediction, visualization of results, and comparison of predictions to experimental structures.

2 EVcouplings framework

The EVcouplings framework integrates the functionality of the previously published methods EVfold (Hopf ; Marks ), EVcomplex (Hopf ) and EVmutation (Hopf ). It provides (i) an easy-to-use command-line application and (ii) a modular Python package containing all functions, data structures and pipelines that comprise the application.

Command-line application: The command-line application allows users to obtain predictions for their proteins and complexes of interest by running the respective EVcouplings pipelines (Fig. 1). Each pipeline is comprised of a series of modular stages that can be configured using a YAML file, which aids reproducibility by documenting all parameters. The pipelines are parallelized and support local multi-process execution as well as commonly used cluster systems, and automatically handles job submission and monitoring. The steps of the prediction pipelines are: , which generates and processes sequence alignments,, which pairs putatively interacting sequences for the protein complex pipeline, which calculates ECs, compare, which compares ECs to experimental structures, , which predicts the effects of mutations, and , which generates de novo 3D models.

Fig. 1.

The EVcouplings Python framework. (a) The protein monomer EVcouplings pipeline entails multiple sequence alignment generation (align stage), EC inference (couplings stage), de novo folding (fold stage), mutation effect prediction (mutate stage) and comparison to experimental structure (compare stage). (b) The protein complex pipeline extends the monomer pipeline to protein interactions by pairing putatively interacting homologs (concatenate stage) and providing restraints for molecular docking (dock stage)

EVcouplings Python package: The command-line application is built on the underlying evcouplings Python package, whose modular architecture and comprehensive documentation facilitate the development of new stages and pipelines. Additionally, the package serves as a toolbox for handling and analyzing EC-related data. Examples for interactive usage are provided in Jupyter notebooks (Kluyver ) distributed with the package, and extensive documentation is available on the web (http://evcouplings.readthedocs.io).

3 Conclusion

EVcouplings is an open source, integrated pipeline for evolutionary couplings analyses. The underlying API serves as a modular basis for data analysis and will allow developers to rapidly create new workflows.