Edit model card

This model is superseded by https://github.com/ORNL/affinity_pred

jglaser/protein-ligand-mlp-1

This is a sentence-transformers model: It maps pairs of protein and chemical sequences (canonical SMILES) onto binding affinities (pIC50 values).

Each member of the ensemble has been trained using a different seed and you can use the different models as independent samples to estimate the uncertainty.

Usage (Sentence-Transformers)

Using this model becomes easy when you have sentence-transformers installed:

#pip install -U sentence-transformers
pip install git+https://github.com/jglaser/sentence-transformers.git@enable_mixed

Then you can use the model like this:

from sentence_transformers import SentenceTransformer
sentences = [{'protein': ["SEQVENCE"], 'ligand': ["c1ccccc1"]}]

model = SentenceTransformer('jglaser/protein-ligand-mlp-1')
embeddings = model.encode(sentences)
print(embeddings)

Evaluation Results

Full Model Architecture

SentenceTransformer(
  (0): Asym(
    (protein-0): Transformer({'max_seq_length': 2048, 'do_lower_case': False}) with Transformer model: BertModel 
    (protein-1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
    (protein-2): Dense({'in_features': 1024, 'out_features': 1024, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
    (ligand-0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (ligand-1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
    (ligand-2): Dense({'in_features': 768, 'out_features': 768, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
  )
  (1): Dense({'in_features': 1792, 'out_features': 1000, 'bias': True, 'activation_function': 'torch.nn.modules.activation.GELU'})
  (2): Dense({'in_features': 1000, 'out_features': 1000, 'bias': True, 'activation_function': 'torch.nn.modules.activation.GELU'})
  (3): Dense({'in_features': 1000, 'out_features': 1000, 'bias': True, 'activation_function': 'torch.nn.modules.activation.GELU'})
  (4): Dense({'in_features': 1000, 'out_features': 1, 'bias': True, 'activation_function': 'torch.nn.modules.linear.Identity'})
  (5): Dense({'in_features': 1, 'out_features': 1, 'bias': True, 'activation_function': 'torch.nn.modules.linear.Identity'})
)

Citing & Authors

Find more information in our bioRxiv preprint

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.