openkinome / kinoml

Structure-informed machine learning for kinase modeling
https://openkinome.org/kinoml/
MIT License
51 stars 21 forks source link

Add skel #1

Closed jaimergp closed 5 years ago

jaimergp commented 5 years ago

Description

Let's try to design a good structure for the package. I have taken some ideas from inspiremd/kinomodel, but it's basically a work in progress.

Currently, the tree is:

kinoml/
├── __init__.py
├── _version.py
├── analysis
│   └── __init__.py
├── data
│   ├── README.md
│   └── look_and_say.dat
├── datasets
│   └── __init__.py
├── docking
│   └── __init__.py
├── features
│   └── __init__.py
├── ml
│   └── __init__.py
├── modeling
│   └── __init__.py
├── optimize
│   └── __init__.py
├── tests
│   ├── __init__.py
│   └── test_kinoml.py
└── workflows
    └── __init__.py

Pinging @jchodera and @jiayeguo for feedback!

jiayeguo commented 5 years ago

Hey Jaime! Thanks for putting this together---it looks good! I have a question though: what's the major function of the optimize module? I guess can fork this gradually add to the featurization part.

jaimergp commented 5 years ago

Hey! Thanks for the feedback!

The idea is that optimize will contain code to relax and minimize protein and protein-ligand structures. First it will use Rosetta-derived routines, but we are aiming at having an openmm-based equivalent.

jchodera commented 5 years ago

What if we call api -> workflows? features could include protein, ligand, and interactions? Otherwise, looks great!