This tool is an inverse folding design model derived from the ProteinMPNN training framework for sequence redesign of acid-tolerant proteins. The training dataset was constructed from extracellular proteins carrying signal peptides from acidophilic microorganisms, which helps avoid the noise introduced by intracellular neutral proteins and allows the model to learn the sequence and physicochemical features of naturally acid-stable secreted proteins. Given a protein PDB backbone, the model outputs candidate amino acid sequences biased toward acid-adaptive features.
This is a computational design tool, and the output sequences still require wet-lab validation for actual acid tolerance. Users only need to upload the backbone structure of the target protein to obtain acid-optimized candidate sequences and accelerate their R&D work.
Upload a protein structure file in PDB or mmCIF format, and the system will automatically parse chain information. After selecting chains to design and setting parameters, the model will generate multiple optimized amino acid sequences, with scores (score = -log_prob, lower is better) and sequence recovery rates for each.