Binder Scoring GUI

A lightweight PySide6 desktop GUI for scoring protein binder structures with several external structural analysis tools:

• FoldX + Aggrescan3D
• PISA
• RosettaDock
• Weighted total score calculation

The GUI is designed for Rocky Linux workstation/server environments and focuses on batch scoring .pdb, .cif, and .mmcif structure files.

---

Table of Contents

• Installation
• System Requirements
• First-run Configuration
• Running the App
• Input and Output
• Scoring Workflow
• Project Layout
• Acknowledgements

---

Installation

Clone or upload this project directory to your Rocky Linux machine.

Create a virtual environment and install Python dependencies:

cd binder_score_gui
python3 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txt

If your system does not already provide virtualenv support:

sudo dnf install -y python3 python3-pip python3-virtualenv

---

System Requirements

This project depends on external tools that are not installed through pip.

Required external programs:

• FoldX
• Aggrescan3D standalone command: aggrescan
• PDBe-KB PISA REST access
• RosettaDock binaries:
  • docking_prepack_protocol
  • docking_protocol
• Rosetta database

Python packages are listed in:

requirements.txt

Current Python dependencies:

• PySide6
• Biopython
• requests

---

First-run Configuration

Open the GUI and click Settings.

Configure the following paths/commands:

FoldX

• FoldX
  • FoldX executable used for RepairPDB and Stability

FoldX + Aggrescan3D

• FoldX4 (A3D)
  • FoldX4 path used by local Aggrescan3D
• A3D Activation
  • shell command used to activate the Aggrescan3D environment
  • example:

module load miniforge3/25.3.0-3 && conda activate a3d

RosettaDock

• Docking Prepack Protocol
  • path to docking_prepack_protocol
• Docking Protocol
  • path to docking_protocol
• Rosetta DB
  • path to the Rosetta database directory

Settings are stored locally in:

resources/settings.json

This file is intentionally ignored by Git because it contains machine-specific paths.

---

Running the App

From the project root:

source .venv/bin/activate
python main.py

If Qt reports DBus warnings in SSH/X11 or non-GNOME sessions, the app may still work normally. If the GUI does not open, try:

dbus-run-session -- python main.py

---

Input and Output

Input directory:

• .pdb
• .cif
• .mmcif

The GUI requires:

• Target chain
• Design chain
• At least one selected scoring program

Default output:

<input_directory>/results.csv

The output file is tab-delimited even though the extension is .csv, so it opens cleanly in Excel.

Runtime intermediate files are written under:

_binder_score_gui/

---

Scoring Workflow

FoldX + Aggrescan3D

Uses the design chain only.

For each input structure:

1. Convert .cif/.mmcif to PDB if needed.
2. Extract the design chain.
3. Run FoldX:

foldx --command=RepairPDB --pdb=<design_chain.pdb>
foldx --command=Stability --pdb=<repaired_pdb>

4. Parse the FoldX Stability .fxout score.
5. Run local Aggrescan3D:

aggrescan -i <design_chain.pdb> -w <work_dir> -D 10 -f <foldx4_path> -v 1

6. Calculate A3D score as the sum of the score column in A3D.csv.

FoldX and Aggrescan3D jobs are run in parallel with a maximum concurrency of 5.

PISA

Uses target + design chains.

For each input structure:

1. Convert .cif/.mmcif to PDB if needed.
2. Extract target and design chains.
3. Submit PDB files to the PDBe-KB PISA REST API in batches.
4. Use the interface with the lowest int_solv_energy.

Reported columns:

• Interface area
• Interface solvation E
• p-value

PISA jobs are submitted with concurrency 5.

RosettaDock

Uses target + design chains.

For each input structure:

1. Convert .cif/.mmcif to PDB if needed.
2. Extract target and design chains.
3. Run prepack:

docking_prepack_protocol -s <input.pdb> -docking:partners <target>_<design>

4. Run 20 parallel docking jobs.
5. Each docking job runs:

docking_protocol -s <prepack_pdb> -partners <target>_<design> -dock_pert 3 8 -nstruct 50

This produces 1000 total docking structures per input.

Reported columns:

• RosettaDock score
• Interface score

The best RosettaDock result is selected by minimum total_score; the matching I_sc is reported as Interface score.

Total Score

Total score is calculated only when all required metrics are available:

• FoldX score
• A3D score
• Interface area
• Interface solvation E
• RosettaDock score
• Interface score

Direction:

• Interface area: higher is better
• All other metrics: lower is better

Weights:

• FoldX score: 0.1
• A3D score: 0.1
• Interface area: 0.1
• Interface solvation E: 0.1
• RosettaDock score: 0.4
• Interface score: 0.2

Reference statistics are stored in:

resources/total_score_reference.json

The current score scale is calibrated so the reference dataset roughly spans 0 to 100. Future values may go below 0 or above 100 if they fall outside the reference range.

---

Project Layout

binder_score_gui/
  app/
    main_window.py
    services/
      processor.py
    widgets/
  resources/
    random_cats/
    total_score_reference.json
    reference_stats.json
  scripts/
    pisa_batch_dir_to_csv.py
    a3d_batch_rest_async.py
  main.py
  README.md
  requirements.txt

---

Acknowledgements

This GUI is a thin interface layer around external tools and services including:

• FoldX
• Aggrescan3D
• PDBe-KB PISA
• RosettaDock

Please follow the licenses, citation policies, and usage requirements of each upstream project.

Return to top