About Alphafold Analysis

af-analysis is a python package for the analysis of AlphaFold protein structure predictions. This package is designed to simplify and streamline the process of working with protein structures generated by AlphaFold 2, AlphaFold 3 and its derivatives like ColabFold, AlphaFold-Multimer and AlphaPulldown.

• Source code repository: https://github.com/samuelmurail/af_analysis

Statement of Need

AlphaFold 2 and its derivatives have revolutionized protein structure prediction, achieving remarkable accuracy. Analyzing the abundance of resulting structural models can be challenging and time-consuming. Existing tools often require separate scripts for calculating various quality metrics (pDockQ, pDockQ2, LIS score) and assessing model diversity. af-analysis addresses these challenges by providing a unified and user-friendly framework for in-depth analysis of AlphaFold 2 results.

Main features:

• Import AlphaFold or ColabFold prediction directories as pandas DataFrames for efficient data handling.
• Calculate and add additional structural quality metrics to the DataFrame, including:
  • pDockQ
  • pDockQ2
  • LIS score
• Visualize predicted protein models.
• Cluster generated models to identify diverse conformations.
• Select the best models based on defined criteria.
• Add your custom metrics to the DataFrame for further analysis.

Installation

• af-analysis is available on PyPI and can be installed using pip:

pip install af_analysis

• You can install last version from the github repo:

pip install git+https://github.com/samuelmurail/af_analysis.git@main

• AF-Analysis can also be installed easily through github:

git clone https://github.com/samuelmurail/af_analysis
cd af_analysis
pip install .

Documentation

The full documentation is available at ReadTheDocs.

Usage

Importing data

Create the Data object, giving the path of the directory containing the results of the alphafold2/colabfold run.

import af_analysis
my_data = af_analysis.Data('MY_AF_RESULTS_DIR')

Extracted data are available in the df attribute of the Data object.

my_data.df

Analysis

• The analysis package contains several function to add metrics like pdockQ and pdockQ2:

from af_analysis import analysis
analysis.pdockq(my_data)
analysis.pdockq2(my_data)

Docking Analysis

• The docking package contains several function to add metrics like LIS Score:

from af_analysis import docking
docking.LIS_pep(my_data)

Plots

• At first approach the user can visualize the pLDDT, PAE matrix and the model scores. The show_info() function displays the scores of the models, as well as the pLDDT plot and PAE matrix in a interactive way.

• plot msa, plddt and PAE:

my_data.plot_msa()
my_data.plot_plddt([0,1])
best_model_index = my_data.df['ranking_confidence'].idxmax()
my_data.plot_pae(best_model_index)

• show 3D structure (nglview package required):

my_data.show_3d(my_data.df['ranking_confidence'].idxmax())

Dependencies

af_analysis requires the following dependencies:

• pdb_numpy
• pandas
• numpy
• tqdm
• seaborn
• cmcrameri
• nglview
• ipywidgets
• mdanalysis

Contributing

af-analysis is an open-source project and contributions are welcome. If you find a bug or have a feature request, please open an issue on the GitHub repository at https://github.com/samuelmurail/af_analysis. If you would like to contribute code, please fork the repository and submit a pull request.

Authors

• Alaa Regei, Graduate Student - Université Paris Cité.
• Samuel Murail, Associate Professor - Université Paris Cité, CMPLI.

See also the list of contributors who participated in this project.

License

This project is licensed under the GNU General Public License version 2 - see the LICENSE file for details.

References

• Jumper et al. Nature (2021) doi: 10.1038/s41586-021-03819-2
• Abramson et al. Nature (2024) doi: 10.1038/s41586-024-07487-w
• Mirdita et al. Nature Methods (2022) doi: 10.1038/s41592-022-01488-1
• Evans et al. bioRxiv (2021) doi: 10.1101/2021.10.04.463034
• Bryant et al. Nat. Commun. (2022) doi: 10.1038/s41467-022-28865-w
• Zhu et al. Bioinformatics (2023) doi: 10.1093/bioinformatics/btad424
• Kim et al. bioRxiv (2024) doi: 10.1101/2024.02.19.580970
• Yu et al. Bioinformatics (2023) doi: 10.1093/bioinformatics/btac749