Introduction
Stack-Tox is an advanced computational tool designed for high-throughput toxicity screening of small molecules. By leveraging ensemble machine learning models trained on extensive chemical databases, it provides reliable predictions regarding a molecule's toxic potential.
Key Feature
Stack-Tox integrates 5 major drug-likeness rule sets (Lipinski, Ghose, Veber, Egan, Muegge) alongside ML toxicity prediction for comprehensive lead profiling.
Getting Started
To begin an analysis, you need a set of chemical structures. Stack-Tox processes these structures on our remote HPC cluster and returns the results asynchronously.
Input Formats
We support the following input methods:
- SMILES Strings: Standard Simplified Molecular Input Line Entry System strings.
- .smi Files: Text files containing one SMILES string per line.
- .txt Files: Plain text files with SMILES.
Example Input Data:
CCO c1ccccc1 CC(=O)Oc1ccccc1C(=O)O
Rules Explained
Beyond toxicity, we evaluate molecules against standard medicinal chemistry filters.
Lipinski (Ro5)
Evaluates oral bioavailability. Checks MW < 500, LogP < 5, HBD < 5, HBA < 10.
Ghose Filter
Qualifies drug-likeness based on physicochemical property ranges (MW, LogP, MR, Atoms).
Veber Rules
Focuses on rotatable bonds (≤ 10) and TPSA (≤ 140) for oral bioavailability.
BOILED-Egg
A graphical model predicting Gastrointestinal Absorption (GIA) and Blood-Brain Barrier (BBB) permeation.
Interpreting Results
Results are color-coded for quick decision making.
API Reference
Stack-Tox provides a RESTful API for integrating toxicity screening into your own pipelines.
Submit Job
/backend/submit_job.php
Parameters:
smiles(string): The SMILES string or list of strings.job_id(int): Unique client-side generated timestamp.
Check Status
/backend/check_status.php
{
"status": "completed",
"data": [
"SMILES,Toxicity,Confidence...",
"CCO,Non-Toxic,0.12..."
]
}