Protein Engineering

AI Scientist integrates state-of-the-art protein engineering tools that bridge the gap between sequence and structure. These tools enable you to predict protein structures, retrieve existing predictions, and design novel protein sequences — all within your research conversation.

ESMFold Protein Folding

Predict 3D protein structures directly from amino acid sequences using ESMFold. This tool is useful for:

Visualizing the fold of a newly designed protein variant
Checking whether mutations disrupt the overall protein structure
Generating structural models for proteins not yet in structural databases

Results are displayed as interactive 3D structure viewers that you can rotate, zoom, and inspect residue-by-residue.

AlphaFold Structure Retrieval

Fetch predicted structures from the AlphaFold database for proteins that already have precomputed models. This provides high-confidence structural predictions for a large portion of known protein sequences without requiring additional computation.

Inverse Folding

Design amino acid sequences that fold into a target 3D structure. Inverse folding is a powerful tool for:

Redesigning a protein to improve stability or solubility while preserving the original fold
Exploring sequence space for a given structural scaffold
Creating de novo proteins with a desired structural topology

Protein Generation

Generate entirely novel protein sequences using AI models. This capability enables exploration beyond the natural sequence space, generating candidates with properties not found in existing proteins.

Guided Protein Generation

Design proteins with specific constraint conditions, such as:

Maintaining key active-site residues while diversifying the surrounding scaffold
Targeting a specific molecular weight or isoelectric point range
Preserving binding interfaces while optimizing other regions for expression or stability

This guided approach combines the creativity of AI-driven generation with the precision of researcher-defined constraints.