mkt is a Python package to generate sequence and structure-based representations for human kinase property prediction. While our application uses this data to predict the impact of clinically observed missense mutations on human kinase activity, we note that many of these tools can be used more extensively to characterize wild-type human kinases and any mutant forms. The databases sub-package can be used to query the APIs of a variety of protein resources that are not exclusive to either kinases or humans, including UniProt, Pfam, and cBioPortal.
Additional documentation can be found here.
mkt is structured as a monorepo with sub-packages described below for specific tasks.
| Subpackages | Description |
|---|---|
schema |
Harmonized and pre-processed sequence and structure data along with Pydantic models to load, query, and validate this data |
databases |
Package containing API clients and scrapers to collect and harmonize kinase data from various sources and generate schema objects |
ml |
In-progress package to build machine learning models to predict kinase properties |
experiments |
In-progress package to analyze experimental results for project |
Sub-packages can be installed directly from Github via pip using the following:
pip install git+https://github.com/choderalab/missense-kinase-toolkit.git#subdirectory=missense_kinase_toolkit/<sub-package directory>
Copyright (c) 2024, Jess White
We would like to express gratitude to the creators of the following resources on which we heavily rely:
Project based on the Computational Molecular Science Python Cookiecutter version 1.1.