Edit model card

PepMLM: Target Sequence-Conditioned Generation of Peptide Binders via Masked Language Modeling image/png In this work, we introduce PepMLM, a purely target sequence-conditioned de novo generator of linear peptide binders. By employing a novel masking strategy that uniquely positions cognate peptide sequences at the terminus of target protein sequences, PepMLM tasks the state-of-the-art ESM-2 pLM to fully reconstruct the binder region, achieving low perplexities matching or improving upon previously-validated peptide-protein sequence pairs. After successful in silico benchmarking with AlphaFold-Multimer, we experimentally verify PepMLM’s efficacy via fusion of model-derived peptides to E3 ubiquitin ligase domains, demonstrating endogenous degradation of target substrates in cellular models. In total, PepMLM enables the generative design of candidate binders to any target protein, without the requirement of target structure, empowering downstream programmable proteome editing applications.

  • Demo: HuggingFace Space Demo Link.[Temporarily Unavailable]
  • Colab Notebook: Link
  • Preprint: Link

Apply for Access

As of February 2024, the model has been gated on HuggingFace. If you wish to use our model, please visit our page on the HuggingFace site (Link) and submit your access request there. An active HuggingFace account is necessary for both the application and subsequent modeling use. Approval of requests may take a few days, as we are a small lab with a manual approval process.

Once your request is approved, you will need your personal access token to begin using this notebook. We appreciate your understanding.

# Load model directly
from transformers import AutoTokenizer, AutoModelForMaskedLM

tokenizer = AutoTokenizer.from_pretrained("TianlaiChen/PepMLM-650M")
model = AutoModelForMaskedLM.from_pretrained("TianlaiChen/PepMLM-650M")

Logo

Downloads last month
1,046
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.

Space using ChatterjeeLab/PepMLM-650M 1