protfunc

Python SDK for the POOL / GRASP-Func protein functional site prediction server, developed by the Ondrechen Research Group at Northeastern University.

POOL (Partial Order Optimum Likelihood) identifies functionally important residues in proteins. GRASP-Func (Graph Representation of Active Sites for the Prediction of Function) classifies protein function by comparing local active-site structures.

Installation

pip install protfunc

Quick Start

from protfunc import Client

client = Client()

# Run POOL on a directory of .pdb files
results = client.run_pool("./structures/", output_dir="./pool_output/")

Usage

POOL

Provide a directory of .pdb files (or a pre-made .zip):

from protfunc import Client

client = Client()
results = client.run_pool("./my_pdb_files/")
# results is a Path pointing to the extracted output directory

GRASP-Func — Step by Step

Run each stage independently. The output of each step feeds into the next:

# Step 1: Pre-processing
preprocessed = client.preprocess("./input/", output_dir="./pre/")

# Step 2: Processing (graph matching)
processed = client.process(preprocessed, output_dir="./proc/")

# Step 3: Visualization
viz = client.visualize(processed, known_proteins=["1a2b", "3c4d"], output_dir="./viz/")

GRASP-Func — Full Pipeline

Or run everything in a single call:

results = client.run_graspfunc(
    "./input/",
    known_proteins=["1a2b", "3c4d"],
    output_dir="./results/",
)

Extract Protein Names

Pull the list of proteins from result files (useful for populating the known_proteins argument):

proteins = client.extract_proteins("./processed/")
# ['protein_a', 'protein_b', ...]

Server Health Check

if client.ping():
    print("Server is up")

Progress Callbacks

By default, the SDK prints progress to the console:

[1/3] preprocess: starting...
[1/3] preprocess: uploading (2.3 MB)...
[1/3] preprocess: running...
[1/3] preprocess: ✓ done (14.2s)
[2/3] process: starting...
...

Custom Callback

Receive structured ProgressEvent objects:

from protfunc import Client, ProgressEvent

def my_callback(event: ProgressEvent):
    print(f"{event.stage} → {event.status.value} ({event.elapsed:.1f}s)")

client = Client(on_progress=my_callback)
client.run_pool("./structures/")

Silent Mode

client = Client(base_url="...", on_progress=False)

Per-Call Override

client.run_pool("./structures/", on_progress=my_callback)
client.preprocess("./input/", on_progress=False)  # silent for this call only

Input Formats

Configuration

Citations

If you use this tool in your research, please cite:

Somarowthu, S. & Ondrechen, M. J. (2012). POOL server: machine learning application for functional site prediction in proteins. Bioinformatics, 28(15), 2078–2079.

Mills, C. L., Garg, R., Lee, J. S., Tian, L., Suciu, A., Cooperman, G. D., Beuning, P. J., & Ondrechen, M. J. (2018). Functional classification of protein structures by local structure matching in graph representation. Protein Science, 27(6), 1125–1135.

License

MIT