Search-ML Documentation

v1.3.1 Linux / macOS / WSL MIT License

Search-ML is a streamlined, serial execution pipeline designed to automate the virtual screening of protein-ligand complexes. It integrates structural preparation, feature extraction using AmberTools and RDKit, and machine learning predictions into a single, unified command.

Download Standalone

🚀 Key Features

Automated Pre-processing

Automatically extracts coordinates, cleans PDBs, and splits chains without manual intervention.

AmberTools Integration

Seamlessly runs tleap for system parameterization and topology generation.

Smart Feature Extraction

Calculates SASA, pocket volume, and electrostatic properties using robust Python modules.

Targeted Screening

Optimized for rapid screening against DrugBank and FDA libraries.

🛠️ Prerequisites

OS: Linux, macOS, or Windows (via WSL2)
Manager: Miniconda or Anaconda installed
Version Control: Git

📥 Installation Guide

Clone the Repository

Terminal

git clone https://github.com/dheeraj-scfbio/search-ml.git
cd search-ml

Download the Model

Download Model (800MB)

# Move downloaded file to the models directory
mv ~/Downloads/final_rf_model.joblib ./models/

Run Installer & Activate

chmod +x install.sh
./install.sh

# 1. Deactivate the conda or other activated environment
conda deactivate

# 2. Check if the variable is set
echo $SEARCH_ML_HOME

# 3. IF the output is empty, run the following to set it:
conda env config vars set SEARCH_ML_HOME="$(pwd)" --name search-ml

# 4. Reload environment to apply changes
conda activate search-ml

🪟 Windows Users (WSL2)

Note: AmberTools requires a Unix-based environment. You must use WSL2 (Windows Subsystem for Linux) to run this tool on Windows.

Step 1: Enable WSL2

Open PowerShell as Administrator.
Run: wsl --install
Restart your computer.
Launch "Ubuntu" from the Start Menu and create a user account.

Step 2: Setup inside Ubuntu

sudo apt-get update && sudo apt-get install git -y
git clone "https://github.com/dheeraj-scfbio/search-ml.git"
cd search-ml

📖 Usage & Syntax

Execution Warning

Do NOT use sh to run the script. Use bash or execute directly.

Syntax

./master.sh

Argument	Description	Example
PDB_FILENAME	Filename with extension	1abc.pdb
LIGAND_CODE	3-letter Residue Name	LIG, MOL
DATABASE	Target Library	DB, FDA

Step-by-Step Examples

Scenario A: DrugBank Screening

Screening protein 9kte.pdb containing ligand 9UM against the DrugBank database.

bash master.sh 9kte.pdb 9UM DB

Scenario B: FDA Screening

Target: target_prot.pdb, Ligand: LIG against approved drugs.

./master.sh target_prot.pdb LIG FDA

Output Explanation

The job creates a directory named after the current timestamp (e.g., 1734177600) to prevent overwriting.

results.txt

The main file containing predicted affinity scores sorted by rank.

job.log

Execution log capturing all STDOUT and STDERR. Check this first if a job fails.

PROT_LIG.pdb

The cleaned and merged complex file used for calculation.

leap.log

Detailed AmberTools log from the tleap process.

Workflow Logic

1
Validation

Checks input PDB existence and validates protein chain length (> 24 AA).
2
Preparation

Splits PDB into PROT.pdb and ligand, then merges into a clean complex.
3
Parameterization

Runs tleap to generate topology and check for missing atoms.
4
Featurization & Prediction

Calculates SASA/Volume via RDKit and feeds data into the Random Forest Regressor.

Troubleshooting

Syntax error near unexpected token

Cause: Running with sh instead of bash.
Fix: Use bash master.sh ...

Protein size must be >= 25 amino acids

Cause: PDB fragment is too small.
Fix: Use a larger structure or check for missing residues.

Maintainer: Dheeraj Kumar Chaurasia

SCFBio, IIT Delhi

Contact Support