Developer notes#
Source codes structure:
FragBEST-Myo
|--- config # .yaml file for train, test, inference (command line usage)
|--- dataset # example and reference files
|--- docs # source of documentation
|--- imgs # source of images
|--- tutorial # tutorial (user guide) to help you to understand how to run our tool
|--- utils # main python codebase
|--- datasets # high-level to deal with dataset
|--- default_config # default configuration to define third-party path
|--- parallel # parallel framework for multiple cores
|--- ppseg # Protein Pocket SEGmentation codebase
|--- analysis # post-process analysis for DL results
|--- holo_descriptor # holo-descriptor related functions
|--- ignite # PyTorch Ignite (DL model train/val/test)
|--- myo # FragBEST-Myo model and auxiliary files
|--- visualization # visualization/plotting of DL results
|--- pymol_scripts # generate PyMOL visualization state (.pse)
|--- tests # testing scripts
|--- thirdparty # script modified from third-party open sources
|--- pytest.ini # pytest configuration
|--- seed.py # fix seed to make the work reproducible
|--- main.py # command line usage script
# environment
|--- build.sh # script for building the environment
|--- pyproject.toml # required packages and functions in pixi environment
# others
|--- .gitignore # git ignore manager
|--- .gitmodules # git submodule manager
|--- .gitattributes # git attribute manager
The below codes should be run in the pixi environment.
# Activate the virtual environment (pixi)
pixi run shell
# in the virtual environment, it looks like:
# (FragBEST-Myo) username@servername:~/path/to/FragBEST-Myo$
Check code style#
poe style-check
Fix code style#
poe style-fix
Test packages#
poe test
Build docs (locally, interactively)#
poe build-doc-dev
Build docs (HTML)#
poe build-doc
Deep learning - Semantic Segmentation (pytorch-ignite)#
The deep learning model framework is based on PyTorch Ignite. We used the codes from the code generator as the starting point and then optimize the workflow.