Unlock Advanced Protein Engineering with ColabDesign on Vecura

What is ColabDesign?

ColabDesign is a unified JAX-based toolkit designed to simplify and accelerate the computational protein engineering process. It seamlessly integrates powerful, state-of-the-art models—including ProteinMPNN for inverse folding and AfDesign for structure-based sequence optimization—behind a single, intuitive API. By enabling users to design sequences from backbones or hallucinate novel proteins from scratch, it bridges the gap between complex algorithmic research and practical laboratory application.

It helps users navigate the protein design landscape by combining structure prediction, sequence generation, and multi-protocol optimization in one place. It is especially useful for researchers who need to generate diverse, foldable protein sequences for scaffolding or functional engineering without needing to manage the underlying heavy technical infrastructure.

What can users do with ColabDesign on Vecura?

With ColabDesign on Vecura, users can:

• Sample diverse sequences for a fixed structural scaffold using the ProteinMPNN backend.
• Optimize seed sequences to fold into a specific target backbone using the AfDesign fixbb protocol.
• Hallucinate novel protein structures de novo by simply specifying a target length and protein constraints.
• Access multi-model structural validations, including pLDDT scores, PAE, and pTM, to assess the confidence of generated protein designs.

What the output means

The output provides comprehensive data including optimized amino-acid sequences, AlphaFold-predicted PDB structures, and key quality metrics such as pLDDT, predicted aligned error (PAE), and pTM scores. Each result includes a curated "best" model selection based on structural confidence metrics, allowing users to easily rank and visualize their designs.

This output should be used to support scientific decision making. It does not replace experimental validation.

Why this matters

The ability to accurately design proteins—choosing amino acids that form stable, functional 3D shapes—is the cornerstone of modern biotechnology. Historically, this required deep expertise in both protein biophysics and large-scale computing. ColabDesign democratizes this process, enabling researchers to perform advanced, gradient-based design optimizations in a streamlined, accessible environment.

By unifying diverse modeling architectures like AlphaFold and ProteinMPNN, this toolkit reduces the barrier to entry for de novo protein design. It provides a robust pipeline for generating candidate molecules that are highly likely to fold into their intended structures, significantly shortening the development cycle for therapeutics, enzymes, and synthetic biology materials.

• Developed by: Sergey Ovchinnikov and collaborators.
• Source: ColabDesign GitHub Repository
• Reference: Dauparas et al. (Science 2022); Jumper et al. (Nature 2021).