SynGlue: A Generative AI Toolkit for PROTACs Decoding and Designing

SynGlue is a powerful Python-based platform for the generation, analysis, and optimization of PROTACs (Proteolysis Targeting Chimeras), and multitarget molecules. Built for researchers in cheminformatics, structural biology, and drug discovery. SynGlue uses data-driven AI to accelerate the design of small molecules, predict degradation potency (DC₅₀, Dmax), and guide rational linker selection. SynGlue offers an end-to-end suite of tools for generating PROTACs and their prioritising.

---

🚀 Features

• Generative AI Models for PROTAC and multitarget molecule design
• TRIE-based fragment storage algorithm for fast retrieval
• Polypharmacology classification
• REST API integration via SynGlue client
• RDKit-powered cheminformatics utilities

---

Installation

📦 Install API Client

pip install synglue requests

🧩 Clone Repository (Optional)

git clone https://github.com/the-ahuja-lab/SynGlue.git
cd SynGlue

---

🔌 SynGlue API Usage

from synglue import SynGlue

client = SynGlue()

# Health check
print(client.health_check())

---

🚀 Design Workflow

# Submit design job
design_result = client.submit_design(target="EGFR", threshold=80)
print(design_result)

# Check status
job_id = design_result["job_id"]
print(client.design_status(job_id))

# client.download_design(job_id, "design_results.zip")

---

🔍 Target Mapping Workflow (List)

molecules = [
    {"name": "Aspirin", "smiles": "CC(=O)Oc1ccccc1C(=O)O"},
    {"name": "Imatinib", "smiles": "CC1=CC=CC=C1"}
]

screen_result = client.submit_screen(molecules=molecules)
print(screen_result)

job_id = screen_result["job_id"]
print(client.screen_status(job_id))

# client.download_screen(job_id, "screen_results.zip")

---

📂 Target Mapping Workflow (CSV)

csv_path = "query.csv"

screen_result = client.submit_screen_csv(csv_path)
print(screen_result)

job_id = screen_result["job_id"]
print(client.screen_status(job_id))

---

📡 API Endpoints (Client Methods)

🔹 Core

• health_check()

🔹 Design

• submit_design(target, threshold=75.0)
• design_status(job_id)
• download_design(job_id, out_path)

🔹 Screening

• submit_screen(molecules)
• submit_screen_csv(csv_path)
• screen_status(job_id)
• download_screen(job_id, out_path)

---

SynGlue Workflow

SynGlue offers a two-part approach:

• Data-Driven

  • Map input molecules, annotate types and targets, and generate optimized molecules.

• Structure-Guided

  • Use structural inputs (PDB, AlphaFold, Rosetta)
  • Perform warhead mapping and molecule generation

---

📊 Type Analysis

Classification	Type
Monovalent, Monotarget	Type 1
Monovalent, Multitarget	Type 2
Multivalent, Monotarget	Type 3
Multivalent, Multitarget	Type 4

---

Tutorials

Tutorial	Description	Colab Link
PROTAC Generation	Generate PROTAC molecules	Open in Colab
Target Mapping	Screen and map targets	Open in Colab

---

Summary

SynGlue enables:

• PROTAC design and prioritization
• Target mapping and screening
• Integration into automated drug discovery pipelines

---

Dependencies

• Python ≥ 3.7
• requests

pip install requests

---

License

SynGlue is licensed under the MIT License. See LICENSE for more details.