Protein function prediction

A python package for the prediction of protein functions as GO terms to participate in the CAFA-5 competition.

The development and testing of the functionalities were made using python3.8, on a linux ─ubuntu 20.4─ environment.

Install

(optional) Create a virtual environment to install the package.

$ mkdir -p venv && python3 -m venv ./venv && source venv/bin/activate

And install it.

$ python3 -m pip install .

Use

Enter a subshell in the virtual environment with dependencies available:

$ source venv/bin/activate

and run python tasks from inside the subshell.

Usage tutorials

• Tutorial1: Working with proteins and and protein structures.
• Tutorial2: Computing GO terms on the training set.
• Tutorial3: Computing candidate GO terms for model training.
• Tutorial4: Order candidate terms by information accrued.
• Tutorial5: Creating a single model and benchmarking its accuracy.
• Tutorial6: Massive training of models.
• Tutorial7: Massive prediction of protein functions.