Subpocket-based structural fingerprint for kinase pocket comparison
The kissim package offers a novel fingerprinting strategy designed specifically for kinase pockets,
allowing for similarity studies across the structurally covered kinome.
The kinase fingerprint is based on the KLIFS pocket alignment,
which defines 85 pocket residues for all kinase structures.
This enables a residue-by-residue comparison without a computationally expensive alignment step.
The pocket fingerprint consists of 85 concatenated residue fingerprints, each encoding a residue’s spatial and physicochemical properties. The spatial properties describe the residue’s position in relation to the kinase pocket center and important kinase subpockets, i.e. the hinge region, the DFG region, and the front pocket. The physicochemical properties encompass for each residue its size and pharmacophoric features, solvent exposure and side chain orientation.
Take a look at the repository kissim_app for pairwise comparison of all kinases to study kinome-wide similarities.
The kissim package documentation is available here, including installation instructions.
Please open an issue if you have questions or suggestions.
We are looking forward to hearing from you!
This work is published under the MIT license.
Copyright (c) 2019, Volkamer Lab
Have you used kissim in your research or found the tool useful? We'd be very grateful if you cited it using the following:
@article{sydow_2022_jcim,
author = {Sydow, Dominique and Aßmann, Eva and Kooistra, Albert J. and Rippmann, Friedrich and Volkamer, Andrea},
title = {KiSSim: Predicting Off-Targets from Structural Similarities in the Kinome},
journal = {Journal of Chemical Information and Modeling},
volume = {62},
number = {10},
pages = {2600-2616},
year = {2022},
doi = {10.1021/acs.jcim.2c00050}
Volkamer Lab's projects are supported by several public funding sources (for more info see our webpage).
The kissim project is a collaboration between the Volkamer Lab (Dominique Sydow, Eva Aßmann and Andrea Volkamer), Albert Kooistra (University of Copenhagen) and Friedrich Rippmann (Merck).
- Cheminformatics and structural bioinformatics:
opencadd,biopython,biopandas - Data science (PyData stack):
numpy,pandas,scipy,jupyter,ipywidgets - Data visualization:
matplotlib,seaborn,nglview - Continuous integration:
pytest,nbval - Documentation:
sphinx,nbsphinx - Code style:
black-nb
Project is based on the Computational Molecular Science Python Cookiecutter version 1.5.