Code for the A3B3 heterohexamer paper

Code to generate and analyse data used in the "De novo design of parallel and antiparallel A3B3 heterohexameric α-helical barrels" paper.

Authors: Joel J. Chubb, Katherine I. Albanese, Alison Rodger, and Derek N. Woolfson.

Correspondence should be addressed to: D.N.Woolfson@bristol.ac.uk

---

Data Availability

All original data files can be found at the Zenodo repository.

---

Requirements

The code requires some external programs to run and reproduce the results:

• ColabFold for structure prediction (Version 1.5.3)
• USalign for structure alignment (Version 20240319)

The work was done using Python 3.10.9. The following Python packages are required:

• cython==3.0.11
• ampal==1.4.0
• pandas==1.5.3
• seaborn==0.12.2
• matplotlib==3.7.0
• scipy==1.11.4
• numpy==1.26.4
• sklearn==1.3.2

---

Installation

To install external programs, please refer to the provided links. To install the necessary Python packages, run:

pip install cython==3.0.11 ampal==1.4.0 pandas==1.5.3 seaborn==0.12.2 matplotlib==3.7.0 scipy==1.11.4 numpy==1.26.4 sklearn==1.3.2

---

Data Generation

All code needed to reproduce the results and analyse outputs is provided. Follow these steps:

1. Specify an output directory and run the script from this directory to generate structure predictions:

   bash ./scripts/run_ColabFold.sh <OUTPUT_DIR>
   
   # Example:
   # bash ./scripts/run_ColabFold.sh ./data/colabfold

   Note: this will generate a large number of files in OUTPUT_DIR (~ 7,000 files, ~ 0.75 GB).

2. Once predictions are done, analyze the structures and cluster results using the notebook (./notebooks/data_analysis.ipynb).

Note: ColabFold was run on an NVIDIA RTX A5500. Outputs may vary depending on your system's configuration.

---

Data Visualisation

The directory also contains IPython notebooks to create all of the data plots used in the manuscript (./notebooks/MS_figure_code.ipynb) and the supplementary materials (./notebooks/SM_figure_code.ipynb).

Note: the output filepaths are hardcoded for our directory hierarchy. These will need to be changed on a per user basis.