CONSTRUCT is a software tool designed to identify functional and structurally important sites in proteins by detecting amino acid sites evolving under strong purifying selection that cluster together in 3D structure.
🧪 Tested on macOS 15 (Sequoia, Apple Silicon - M3)
- macOS (Apple Silicon — M1/M2/M3)
- Minimum macOS 11 (Big Sur)
- Not compatible with Intel Macs at the moment
- Linux
- Ubuntu 20.04 LTS or later (Debian-based systems supported)
| Tool | Minimum Version | Recommended | Notes |
|---|---|---|---|
| Python | 3.7 | 3.10+ | Required for GUI (customtkinter) |
| R | 4.1.0 | 4.1.2+ | Ensures CRAN binary compatibility |
| Homebrew | — | Latest | Required for macOS dependency handling |
sudo apt install r-base-core python3-pip gitAll of the following dependencies are installed automatically by the script:
customtkinter
tidyversereadrdplyrbio3dmsa(and its Bioconductor dependencies)
rate4site(compiled from source)- Compilation tools (
gcc,make, etc.)
After downloading the files from the repository, run the installer script. It will verify and install all dependencies.
git clone https://github.com/Rcoppee/CONSTRUCT
cd CONSTRUCT/
bash install_packages.shTo launch the program:
python3 CONSTRUCT.pyA graphical interface will open:
Just fill in the necessary fields and click "Run post-processing" to start the analysis.
CONSTRUCT generates three result files:
- spatial_rates.txt: a file containing the spatially correlated site-specific substitution rates of amino acid sites, ranked by their level of conservation.
- log_files.txt: indicates whether a patch of conserved amino acid sites was detected in the protein structure (with the best window size and corresponding correlation strength).
- color_conserved.pml: a file highlighting the top 10% of conserved amino acid sites (for use with PyMOL).
To analyze the KEAP1 propeller domain, two files must be submitted:
- A fasta file: This file should contain an alignment of orthologous sequences with the reference sequence listed first.
- A PDB file: This file should contain the Cartesian coordinates of the protein structure (in this example we hase used the PDB ID: 2FLU).
Once you have submitted these files, you can proceed by running the post-processing tool. When the process is complete, you'll see a score representing the strength of the correlation in site-specific substitution rates (a value > 8 indicates the presence of a patch of conserved amino acid sites). In this example, using the side-chain orientation option as Cartesian coordinates, you might observe a log score of 74.63, which is > 8, indicating the presence of a patch of conserved amino acid sites (corresponding to the surface interface with Nrf2, the substrate of KEAP1).
To visualize this patch, you can use PyMOL:
- Open PyMOL.
- Go to "File" and select "Open."
- Load the generated
color_conserved.pmlfile.
/!\ If you move the PDB file after running CONSTRUCT, you'll have to change the first line of color_conserved.pml, because the first line is: load {pdb_file_path}/my_pdb.pdb (where my_pdb.pdb is your PDB file). You can also manually open the PDB file in PyMOL then open color_conserved.pml.
Let’s take DHPS as an example.
In the initial analysis, no specific boundaries were set, and the following patch was identified:
This patch is located on the DHPS domain of the protein.
If you want to focus on a specific part of the protein, such as the PPPK domain, you can define the boundaries for that domain, which in this case would be from position 1 to 386.
After specifying these boundaries, a patch of conserved amino acid sites was specifically detected in the PPPK domain:
A video tutorial has been created for easy installation and execution of CONSTRUCT: https://www.youtube.com/watch?v=bf-VYReZIeM&t=10s
CONSTRUCT: an algorithmic tool for identifying functional or structurally important regions in protein tertiary structure
Lucas Chivot, Noé Mathieux, Anna Cosson, Antoine Bridier-Nahmias, Loic Favennec, Jean-Christophe Gelly, Jérôme Clain, Romain Coppée