DeProteo

• 

  DeProteo UI and Sample Prompting

Inspiration

We wanted to streamline protein design by combining AI-driven prompt generation with structural validation, making it faster and more intuitive to create functional proteins.

What it does

DeProteo takes a user’s design prompt and generates candidate protein structures, evaluates them against key structural metrics, and outputs validated, ready-to-use designs.

How we built it

We integrated a large language model (LLM) for prompt interpretation with structural evaluation tools, including PDB analysis and interface metrics. The workflow automates design, assessment, and output of protein structures.

Challenges we ran into

Translating natural language prompts into meaningful protein designs was tricky, especially ensuring the outputs are physically plausible and structurally stable. Handling diverse protein topologies and large datasets also required careful optimization.

Accomplishments that we're proud of

We built an end-to-end pipeline that can autonomously generate, validate, and output protein designs. The system successfully bridges AI prompt generation with structural biology evaluation, enabling rapid prototyping.

What we learned

We gained insight into the complexities of protein structure validation, the nuances of translating prompts into designable sequences, and the importance of automating evaluation metrics for reproducibility.

What's next for DeProteo

We plan to expand metrics for functional validation, improve design diversity, and integrate experimental feedback loops, ultimately making DeProteo a robust tool for both computational and lab-based protein engineering.

Built With

• python
• shell