A Unified Deep Learning Model for Multi-task Reaction Predictions.

It is built on huggingface transformers -- T5 model with some modifications.

T5Chem can be either installed via pip or from source. We recommend to install t5chem from source.

1. To install from source (with latest version):

$ git clone https://github.com/HelloJocelynLu/t5chem.git
$ cd t5chem/
$ python setup.py install
$ python setup.py test # optional, only works when you have pytest installed

It should automatically handle dependencies for you.

2. To install via pip

$ pip install t5chem

Call from command line:

$ t5chem -h # show the general help information
$ t5chem train -h # show help information for model training
$ t5chem predict -h # show help information for model prediction

We have some sample data (a small subset from datasets used in paper) available in data/ folder, to have a quick start:

$ tar -xjvf data/sample_data.tar.bz2
$ t5chem train --data_dir data/sample/product/ --output_dir model/ --task_type product --pretrain models/pretrain/simple/ --num_epoch 30
$ t5chem predict --data_dir data/sample/product/ --model_dir model/

Note that we may not get a very good result (~70% top-1 accuracy) as we are only trained on a small dataset.

Required auguments:

• data_dir: The path to data directory, should contain train.source, train.target for training, and test.source, test.target for testing
• output_dir: The directory to save trained model. Will generate four files: vocab.pt for vocabulary, config.json for model configuration, pytorch_model.bin for trained model weights and training_args.bin for training arguments.
• task_type: task type. Currently we are supporting product (for forward reaction prediction), reactants (for single-step retrosynthesis), reagents (for reagents prediction), regression (to predict numeric values) and classification (to predict categorical values).
• model_dir: The path to a trained model. (task type, tokenization method...etc can be inferred from saved model.

Call as an API:

from transformers import T5ForConditionalGeneration
from t5chem import T5ForProperty, SimpleTokenizer
pretrain_path = "path/to/your/pretrained/model/"
model = T5ForConditionalGeneration.from_pretrained(pretrain_path)    # for seq2seq tasks
tokenizer = SimpleTokenizer(vocab_file=os.path.join(pretrain_path, 'vocab.pt'))
inputs = tokenizer.encode("Product:COC(=O)c1cc(COc2ccc(-c3ccccc3OC)cc2)c(C)o1.C1CCOC1>>", return_tensors='pt')
output = model.generate(input_ids=inputs, max_length=300, early_stopping=True)
tokenizer.decode(output[0], skip_special_tokens=True) # "COc1ccccc1-c1ccc(OCc2cc(C(=O)O)oc2C)cc1"

model = T5ForProperty.from_pretrained(pretrain_path)  # for non-seq2seq task
inputs = tokenizer.encode("Classification:COC(=O)c1cccc(C(=O)OC)c1>CN(C)N.Cl.O>COC(=O)c1cccc(C(=O)O)c1", return_tensors='pt')
outputs = model(inputs)
print(outputs.logits.argmax())   # Class 3

We have Google Colab examples available! Feel free to try it out:

• Call T5Chem via CLI (command line) Colab
• Use a pretrained model in python script Colab
• Design your own project: predict molecular weights Colab

• Now we have found some installation issues on rdkit version later than 2020.09.2 (See discussion here)
• torchtext version 0.10.0 published some backward incompatible changes. T5Chem now only tested on torchtext<=0.8.1

MIT Licence.

t5chem was written by Jocelyn Lu.