KinFragLib: Kinase-focused fragment library

Note: This repository is constantly updated, hence the statistics and numbers derive from the paper. The current fragmentation library is based on the KLIFS database downloaded on 06.12.2023. You can retrieve the repository state for the published KinFragLib paper in release v1.0.0.

Table of contents

• Repository content
• Description
• Quick start
• Contact
• License
• Citation
• List of publications

Repository content

This repository holds the following resources:

1. Fragment library data and a link to the combinatorial library data.
2. Quick start notebook explaining how to load and use the fragment library.
   

3. Notebooks

   3.1. KinFragLib: Notebooks covering the full analyses regarding the fragment and combinatorial libraries as described in the corresponding paper.
   3.2. CustomKinFragLib: Notebooks providing a custom filtering framework to reduce the fragment library size.

Please find detailed descriptions of files in data/ and notebooks/ in the folders' README files.

Description

Exploring the kinase inhibitor space using subpocket-focused fragmentation and recombination

Protein kinases play a crucial role in many cell signaling processes, making them one of the most important families of drug targets. Fragment-based drug design has proven useful as one approach to developing novel kinase inhibitors. Usually, fragment-based methods follow a knowledge-driven approach, i.e., optimizing a focused set of fragments into molecular hits.

We present here KinFragLib, a data-driven kinase-focused fragment library based on the structural kinome data retrieved from the KLIFS database. Each kinase binding pocket (for DFG-in structures with non-covalent ligands) is automatically divided in KinFragLib into six subpockets, i.e. the adenine pocket (AP), front pocket (FP), solvent-exposed pocket (SE), gate area (GA) as well as back pocket 1 and 2 (B1 and B2), based on defined pocket-spanning residues. Each co-crystallized ligand is fragmented using the BRICS algorithm and its fragments are assigned to the respective subpocket they occupy. Following this approach, a fragment library is created with respective subpocket pools. This fragment library enables an in-depth analysis of the chemical space of known kinase inhibitors and can be used to enumerate recombined fragments in order to generate novel potential inhibitors.

We have added an extension with CustomKinFragLib which provides a pipeline to filter the fragments in KinFragLib checking for unwanted substructures (PAINS and Brenk et al.), drug-likeness (Rule of Three and QED), synthesizability (similarity to buyable building blocks and SYBA) and pairwise retrosynthesizability. Each filter can be (de-)activated and the parameters can be modified by the user to create a customized filtered fragment library.

Quick start

1. Clone this repository.

   bash git clone https://github.com/volkamerlab/KinFragLib.git

2. Create the kinfraglib conda environment.

   ```bash

   Change to KinFragLib directory

   cd /path/to/KinFragLib

   Create environment

   Hint: if conda is too slow, consider mamba instead

   conda env create -f environment.yml

   When using a MacBook with an M1 chip you may need instead:

   CONDA_SUBDIR=osx-64 conda env create -f environment.yml

   Activate environment

   conda activate kinfraglib

   Install the kinfraglib pip package

   cd .. pip install -e KinFragLib ```

3. Open the notebook quick_start.ipynb for an introduction on how to load and use the fragment library.

   ```bash

   Change to KinFragLib directory (if you have not already)

   cd /path/to/KinFragLib

   Start jupyter lab to explore the notebooks

   jupyter lab ```

Contact

Please contact us if you have questions or suggestions.

• Open an issue on our GitHub repository: https://github.com/volkamerlab/KinFragLib/issues
• Or send us an email: volkamer@cs.uni-saarland.de

We are looking forward to hearing from you!

License

This resource is licensed under the MIT license, a permissive open-source license.

Citation

CustomKinFragLib publication

Kramer, P. L., Buchthal, K., Sydow, D., Leo, K. L., and Volkamer, A. CustomKinFragLib: Filtering the kinase-focused fragmentation library. ChemRxiv preprint. 2025. https://doi.org/10.26434/chemrxiv-2025-3gz92

bib @article{doi:10.26434/chemrxiv-2025-3gz92, author = {Kramer, Paula Linh and Buchthal, Katharina and Sydow, Dominique and Leo, Katharina Sonja and Volkamer, Andrea}, title = {CustomKinFragLib: Filtering the kinase-focused fragmentation library}, journal = {ChemRxiv}, year = {2025}, URL = {https://doi.org/10.26434/chemrxiv-2025-3gz92} }

Original KinFragLib publication

Sydow, D., Schmiel, P., Mortier, J., and Volkamer, A. KinFragLib: Exploring the Kinase Inhibitor Space Using Subpocket-Focused Fragmentation and Recombination. J. Chem. Inf. Model. 2020. https://pubs.acs.org/doi/abs/10.1021/acs.jcim.0c00839

bib @article{doi:10.1021/acs.jcim.0c00839, author = {Sydow, Dominique and Schmiel, Paula and Mortier, Jérémie and Volkamer, Andrea}, title = {KinFragLib: Exploring the Kinase Inhibitor Space Using Subpocket-Focused Fragmentation and Recombination}, journal = {Journal of Chemical Information and Modeling}, volume = {60}, number = {12}, pages = {6081-6094}, year = {2020}, doi = {10.1021/acs.jcim.0c00839}, note ={PMID: 33155465}, URL = {https://doi.org/10.1021/acs.jcim.0c00839} }

List of publications

• Kinase Inhibitor Scaffold Hopping with Deep Learning Approaches Lizhao Hu, Yuyao Yang, Shuangjia Zheng, Jun Xu, Ting Ran, and Hongming Chen Journal of Chemical Information and Modeling 2021 10.1021/acs.jcim.1c00608
• TWN-FS method: A novel fragment screening method for drug discovery Yoon, Hye Ree and Park, Gyoung Jin and Balupuri, Anand and Kang, Nam Sook Computational and Structural Biotechnology Journal 2023 10.1016/j.csbj.2023.09.037
• Efficient Hit-to-Lead Searching of Kinase Inhibitor Chemical Space via Computational Fragment Merging Grigorii V. Andrianov, Wern Juin Gabriel Ong, Ilya Serebriiskii, and John Karanicolas Journal of Chemical Information and Modeling 2021 10.1021/acs.jcim.1c00630
• KiSSim: Predicting Off-Targets from Structural Similarities in the Kinome Dominique Sydow, Eva Aßmann, Albert J. Kooistra, Friedrich Rippmann, and Andrea Volkamer Journal of Chemical Information and Modeling 2022 10.1021/acs.jcim.2c00050
• Target-Focused Library Design by Pocket-Applied Computer Vision and Fragment Deep Generative Linking Merveille Eguida, Christel Schmitt-Valencia, Marcel Hibert, Pascal Villa, and Didier Rognan Journal of Medicinal Chemistry 2022 10.1021/acs.jmedchem.2c00931
• Guided docking as a data generation approach facilitates structure-based machine learning on kinases Backenköhler M, Groß J, Wolf V, Volkamer A. ChemRxiv 2023 10.26434/chemrxiv-2023-prk53 This content is a preprint and has not been peer-reviewed.
• Constructing Innovative Covalent and Noncovalent Compound Libraries: Insights from 3D Protein–Ligand Interactions Xiaohe Xu, Weijie Han, Xiangzhen Ning, Chengdong Zang, Chengcheng Xu, Chen Zeng, Chengtao Pu, Yanmin Zhang, Yadong Chen, and Haichun Liu Journal of Chemical Information and Modeling 202410.1021/acs.jcim.3c01689