FastEnsemble¶

FastEnsemble is a Python-based application that integrates MSA generation, structure prediction via AlphaFold 2 (AF2), and interactive analysis of protein conformational ensembles, all in one place. Uniquely, this tool enables researchers to leverage ML to generate protein conformations and analyze their populations without running MD simulations.

Citation¶

FastEnsemble is based off of the research described in the manuscripts below.

Monteiro da Silva, G., Cui, J.Y., Dalgarno, D.C. et al. High-throughput prediction of protein conformational distributions with subsampled AlphaFold2. Nat Commun 15, 2464 (2024). https://doi.org/10.1038/s41467-024-46715-9

Features¶

MSA Generation: Automatically generates multiple sequence alignments (MSAs) from amino acid sequences using JACKHMMER and MMseqs2.
Structure Prediction: Predicts protein structures using the ColabFold implementation of AlphaFold 2 (AF2).
Conformational Ensembles: Generates alternative protein conformations through MSA subsampling.
Interactive Analysis: Analyzes protein conformational ensembles and the effects of mutations on protein dynamics using our suite of analysis tools.
User-Friendly GUI: Accessible through an intuitive graphical user interface suitable for non-programmers.

Installation¶

To install FastEnsemble, download and run the installation script provided in the repository:

./install.sh

If you wish to install only the CLI version without the gui, run this script instead:

./install_cli.sh

This script will set up the necessary environment and dependencies. At the end of the script we add the source path to the bashrc file so that you can use any of the commands without the need to activate the conda environment.

Next, close the terminal window and reopen a new one.

Now, run any of the commands reported in the Usage section.

Dependencies¶

The installation script sets up the environment and installs all necessary dependencies.

FastEnsemble relies on the following dependencies:

Conda Packages¶

git
python=3.10
openmm==7.7.0
pdbfixer
kalign2=2.04
hhsuite=3.3.0
mmseqs2=15.6f452
hmmer
scikit-learn
mdanalysis
seaborn
scipy

Python Packages (via pip)¶

PyQt5
pandas
pyqt
matplotlib
silence_tensorflow
pyqtgraph
colabfold
silence_tensorflow
pdb-tools

Documentation¶

Documentation can be found on this ReadTheDocs page.

OS Requirements¶

This package has been tested on Linux RedHat7. NOTE: To use ColabFold-dependent features (mmseqs2 msa and AlphaFold2 predictions) use Ubuntu versions older than 20.04.

Usage¶

Before running any commands, download this file and run it:

./fast_ensemble.sh

Running the GUI¶

To start the graphical user interface, execute the following command:

run_gui

This will launch the main application window, where you can access various functionalities such as submitting new jobs, checking job status, and viewing analysis logs.

Running via Command-line¶

First, run

fast_ensemble_init

Next, run any of the following, either specifying a config file path, or by specifying the parameters via command-line arguments.

Sample config files and sample results are available via this link Download Sample Files.

For more detailed instructions on how to use each tool, refer to the ReadTheDocs documentation.

MSA Generation:

jackhmmer_msa: Generate MSA using jackhmmer.

jackhmmer_msa --config_file <path_to_config>

mmseqs2_msa: Generate MSA using mmseqs2.

mmseqs2_msa --config_file <path_to_config>

Prediction:

predict_ensemble: Run ensemble predictions.

predict_ensemble --config_file <path_to_config>

fast_ensemble_init: Initialize sample config file.
```
fast_ensemble_init
```

Analysis:

rmsd_mode1d: Analyze RMSD in 1D mode.

rmsd_mode1d --config_file <path_to_config>

rmsd_mode2d: Analyze RMSD in 2D mode.

rmsd_mode2d --config_file <path_to_config>

tmscore_mode1d: Analyze TM-score in 1D mode.

tmscore_mode1d --config_file <path_to_config>

tmscore_mode2d: Analyze TM-score in 2D mode.

tmscore_mode2d --config_file <path_to_config>

pca_clustering: Perform PCA clustering on the predicted structures.
```
pca_clustering --config_file <path_to_config>
```
rmsf_plddt: Calculate RMSF and pLDDT for the predicted structures.
```
rmsf_plddt --config_file <path_to_config>
```
save_traj: Save trajectories from the analysis.
```
save_traj --config_file <path_to_config>
```

Troubleshooting¶

When running MSA generation or predictions, if you are getting a no such file or directory error, it is likely an issue with the AlphaFold2 Installation. Try reinstalling the package and ensure that the AF2 installation is on path.
For issues with the qt platform, if the ‘xcb’ platform is found but cannot be initialized, try this command, or refer to this github issue

sudo apt-get install libx11-xcb1 libxcb1 libxcb-glx0 \
libxcb-keysyms1 libxcb-image0 libxcb-shm0 libxcb-icccm4 \
libxcb-sync1 libxcb-xfixes0 libxcb-shape0 libxcb-randr0 \
libxcb-render-util0 libxcb-render0 libxcb-xinerama0 libxcb-xkb1 libxkbcommon-x11-0

Download Sample Files¶

To get started quickly, download the sample files from the link below and add them to the root directory of the project:

Download Sample Files

Contributing¶

We welcome contributions to FastEnsemble. If you would like to contribute, please follow these steps:

Fork the repository.
Create a new branch (git checkout -b feature-branch).
Make your changes.
Commit your changes (git commit -m 'Add some feature').
Push to the branch (git push origin feature-branch).
Create a new Pull Request.

Please ensure your code follows our coding standards and includes appropriate tests.

License¶

FastEnsemble is licensed under the MIT License. See the LICENSE file for more details.

Acknowledgements¶

Project based on the Computational Molecular Science Python Cookiecutter version 1.1.