colabfold

msa-search

RUN ANYWHERE

Generates a multiple sequence alignment from a query sequence and a protein sequence database search.

biology bionemo protein folding nim drug discovery

Get API Key

API Reference

Select your target environment

Follow the steps below to download and run the NVIDIA NIM inference microservice for this model on your infrastructure of choice.

Generate API Key

Pull and run colabfold/msa-search using Docker (this will download the full model and run it in your local environment)


$ docker login nvcr.io
Username: $oauthtoken
Password: <PASTE_API_KEY_HERE>

Start the MSA Search NIM

Export the NGC_API_KEY environment variable.


export NGC_API_KEY=<your personal NGC key>

The NIM container automatically downloads any required models. To save time and bandwidth it is recommended to provide a local cache directory. This way the NIM will be able to reuse any already downloaded models. Execute the following command to setup the cache directory:


export LOCAL_NIM_CACHE=~/.cache/nim
mkdir -p $LOCAL_NIM_CACHE

Run the NIM container with the following commands.


docker run -it \
    --runtime=nvidia \
    -p 8000:8000 \
    -e NGC_API_KEY \
    -v "$LOCAL_NIM_CACHE":/opt/nim/.cache \
    nvcr.io/nim/colabfold/msa-search:1.0.0

This will by default run on all available GPUs. Below is an example of running the NIM specifically on device 1:


docker run -it \
    --runtime=nvidia \
    --gpus='"device=0"' \
    -p 8000:8000 \
    -e NGC_API_KEY \
    -v "$LOCAL_NIM_CACHE":/opt/nim/.cache \
    nvcr.io/nim/colabfold/msa-search:1.0.0

Query the NIM

The following python script can be saved to a file named query-msa-search.py and can then be run using python query-msa-search.py. This will post a request to the locally-running NIM and print the response if it returns successfully. If the NIM fails, the response's text field will be printed.


import requests
import json

def main():
    # Define the URL to which the request will be sent
    NIM_URL_FULL_PATH = "http://0.0.0.0:8000/biology/colabfold/msa-search/predict"  # Replace with your actual URL

    # Define the sequence data
    sequence = (
        "MVPSAGQLALFALGIVLAACQALENSTSPLSADPPVAAAVVSHFNDCPDSHTQFCFHGTCRFL"
        "VQEDKPACVCHSGYVGARCEHADLLAVVAASQKKQAITALVVVSIVALAVLIITCVLIHCCQVRKHCEWCR"
        "ALICRHEKPSALLKGRTACCHSETVV"
    )

    # Set the request headers
    headers = {
        "Content-Type": "application/json"
    }

    # Prepare the payload
    data = {
        "sequence": sequence,
        "search_type": "colabfold"
    }

    print("Making request to", NIM_URL_FULL_PATH)

    # Make the POST request
    response = requests.post(NIM_URL_FULL_PATH, headers=headers, data=json.dumps(data))

    # Check the response status
    if response.status_code == 200:
        print("Request was successful.")
        # Optionally, process the response content
        with open('response.json', 'w') as f:
            json.dump(response.json(), f)
    else:
        print(f"Unexpected status code: {response.status_code}")
        print("Response data:", response.text)
if __name__ == "__main__":
    main()