ThermoMPNN Mutation Stability (ΔΔG)

This tool predicts how a mutation changes how stably a protein stays folded. The quantity it reports is ΔΔG, the change in folding free energy, measured in kcal/mol (a unit of energy). The sign tells you the direction: a negative value means the mutation is predicted to stabilize the fold, and a positive value means it destabilizes it.

You upload the protein structure, and then either give a list of specific mutations to score, or ask for a saturation scan, which tries every one of the standard amino acids at each position along a stretch of a chain. It uses ThermoMPNN, a model trained to predict these stability changes.

The minimal input is the structure plus the fixed mode of stability, together with either a list of mutations or a saturation request.

When to use it

Use it when you want to find mutations that make a protein more stable, or to flag ones that would weaken it, before testing anything in the lab. The mutation list mode is for checking specific changes you already have in mind. The saturation scan is for discovery, when you want to see, position by position, which substitutions help and which hurt across a region of interest.

Inputs

Input	Required	What it is
`mode`	yes	Always `stability` for this tool. It is set for you.
`pdb_file`	one of	The protein structure, uploaded as a PDB file. Provide this or `pdb`.
`pdb`	one of	The same structure supplied inline as a JSON object instead of a file. Provide this or `pdb_file`.
`mutations`	one of	A list of specific mutations to score. Provide this or `saturation`.
`saturation`	one of	A request to scan a stretch of a chain and try every standard amino acid at each position. Provide this or `mutations`.

Two further optional fields tune the run: batch_size (how many mutations are scored at once, default 256) and checkpoint (which trained model version to use, default thermompnn_v1). Leave them at their defaults for a first run.

How to run it

Submit your structure from Azulene Studio, the Python SDK, or the CLI. Local file paths are uploaded for you automatically. New here? The Get started page walks through installing, logging in, and running a ready made example first.

In Azulene Studio

Open ThermoMPNN Mutation Stability from the tools list, then on the Inputs and Parameters step upload your protein PDB and either enter the specific mutations you want to score or set up a saturation scan over a chain and range, then Review and Submit.

From the Python SDK

from opal import jobs

result = jobs.submit(
    job_type="mpnn_stability",
    input_data={
        "mode": "stability",
        "pdb_file": "protein.pdb",
        "mutations": ["A:T17V", "A:K22R"],
    },
)

From the CLI

Pass the inputs as a JSON string. File paths are uploaded automatically.

opal jobs submit --job-type mpnn_stability \
  --input-data '{"mode": "stability", "pdb_file": "protein.pdb", "mutations": ["A:T17V", "A:K22R"]}'

Reading the result

You get back a predicted ΔΔG for each mutation, in kcal/mol. Read the sign first: negative values are predicted to stabilize the fold and positive values to destabilize it, and the further from zero, the larger the predicted effect. Rank by the most negative values to find your best stabilizing candidates.

For a mutation list, you get one row per mutation. For a saturation scan, you get a value for every amino acid at every scanned position, which you can read as a map of where on the protein substitutions help or hurt.

Notes

Supply either pdb_file or pdb, and either mutations or saturation. A saturation scan over a long stretch produces many predictions, so it takes longer than a short mutation list; the batch_size default keeps large scans efficient. ThermoMPNN runs on a GPU. For a physics based estimate of the same fold stability question, see Protein Mutation ΔΔG Fold.