---
license: cc-by-nc-nd-4.0
extra_gated_fields:
  Name: text
  Company: text
  Country: country
  Specific date: date_picker
  I want to use this model for:
    type: select
    options:
    - Research
    - Education
    - label: Other
      value: other
  I agree to share generated sequences and associated data with authors before publishing: checkbox
  I agree not to file patents on any sequences generated by this model: checkbox
  I agree to use this model for non-commercial use ONLY: checkbox
base_model:
- facebook/esm2_t30_150M_UR50D
pipeline_tag: fill-mask
---

# MeMDLM: De Novo Membrane Protein Design with Masked Discrete Diffusion Language Models

![image/png](https://cdn-uploads.huggingface.co/production/uploads/65bbea9a26c639b000501321/QQoEGgvJddR2JfE7tzMSR.png)


[Masked Diffusion Language Models (MDLMs)](arxiv.org/pdf/2406.07524), introduced by Sahoo et al, provide strong generative capabilities to BERT-style models. In this work, we pre-train and fine-tune ESM-2-150M protein language model (pLM) on the MDLM objective to scaffold functional motifs and unconditionally generate realistic, high-quality membrane protein sequences.


## Repository Authors
[Shrey Goel](mailto:shrey.goel@duke.edu), Undergraduate Student at Duke University <br>
[Vishrut Thoutam](mailto:vishrut.thoutam64@gmail.com), Student at High Technology High School <br>
[Pranam Chatterjee](mailto:pranam.chatterjee@duke.edu), Assistant Professor at Duke University 

Reach out to us with any questions!