MolBridge

Bridging the Gap Between Molecule and Textual Descriptions via Substructure-aware Alignment

EMNLP 2025 Main Conference

MolBridge is a substructure-aware generative model designed to align molecular structures and textual descriptions.
We release checkpoints for bidirectional generation between molecule representations and natural language captions.

---

🔗 Released Models

---

📦 Released Datasets

All datasets for MolBridge and MolBridge-Gen can be found here: Google Drive.

For the CHEBI-20 dataset, please refer to the original MolT5 repository and documentation.

---

🧪 Usage Examples

Caption → SMILES (C2S)

from transformers import AutoTokenizer, T5ForConditionalGeneration

tokenizer = AutoTokenizer.from_pretrained("laituan245/molt5-base", model_max_length=512)
model = T5ForConditionalGeneration.from_pretrained("PhTae/MolBridge-Gen-Base-C2S")

caption = "The molecule is a monoterpene that is bicyclo[2.2.1]heptane substituted by methyl groups at positions 1, 3 and 3. It is a monoterpene, a terpenoid fundamental parent and a carbobicyclic compound."
caption = "Provide a molecule based on this description: " + caption

token = tokenizer(caption, return_tensors="pt", padding="longest", truncation=True)
gen_results = model.generate(
    input_ids=token["input_ids"],
    attention_mask=token["attention_mask"],
    num_beams=5,
    max_new_tokens=512
)

SMILES → Caption (S2C)

from transformers import AutoTokenizer, T5ForConditionalGeneration

tokenizer = AutoTokenizer.from_pretrained("laituan245/molt5-base", model_max_length=512)
model = T5ForConditionalGeneration.from_pretrained("PhTae/MolBridge-Gen-Base-S2C")

canonicalized_smiles = "CC(=O)N[C@@H](CCCN=C(N)N)C(=O)[O-]"
canonicalized_smiles = "Provide a whole description of this molecule: " + canonicalized_smiles

token = tokenizer(canonicalized_smiles, return_tensors="pt", padding="longest", truncation=True)
gen_results = model.generate(
    input_ids=token["input_ids"],
    attention_mask=token["attention_mask"],
    num_beams=5,
    max_new_tokens=512
)