The Chemistry Development Kit (CDK, research-software-directory.org/software/cdk) (Steinbeck et al. 2003, Steinbeck et al. 2006, Willighagen et al. 2017), rdkit, and OpenBabel are the three leading open source cheminformatics tools, powering much of the cheminformatics research. For example, the Chemistry Development Kit (CDK), such as the generic workflow platform KNIME (Beisken et al. 2013) and PaDEL-descriptor (Yap 2010) used in chemical property prediction research, each cited more than 1000 times. The CDK publications are cited more than two thousand times and the software is mentioned almost 50 thousand times on GitHub.

The most recent CDK release is CDK 2.9 of August 2023 (Mayfield et al. 2023). However, many tools using the CDK use an older version of the CDK. Each CDK release improves the use of open standards, such as SMILES, InChI, and open data, such as the latest IUPAC isotope masses and element names. Use of the latest CDK version improves the interoperability of the tools using it. Our intention is that important CDK-based tools use the latest CDK version, for which we will collaborate with the respective open science projects.

The vision of this project is two-fold. For the first part (see WP1), this project will improve the CDK library by introducing and making it compatible with newer Java programming language features. While developing these, the coverage of JavaDoc and unit testing with JUnit will be extended. Moreover, OSGi support will be improved, where a current limitation is that multiple OSGi bundles contain the same Java package, causing problems with OSGi-based software, like OpenChrom (Wenig and Odermatt 2010) and PathVisio (Kutmon et al. 2015).

The second part of this project will focus on updating tools using the CDK to the latest CDK version (currently 2.9, but later versions when they are released; see WP2). Tools of particular interest for updating are JChemPaint (Krause et al. 2000), AMBIT (Jeliazkova and Jeliazkov 2011), rcdk, OpenChrom, DECIMER (Rajan et al. 2023), KNIME, and PaDEL-descriptor. These downstream tools will need to be updated for changes in the application programming interfaces (APIs) of the CDK. Required API changes will be publicly shared and disseminated with the Groovy Cheminformatics with the Chemistry Development Kit book (egonw.github.io/cdkbook/). The applicants have a long-standing collaboration with the developers of most of the tools of interest.

Finally, we will present the results of this project at an open, international user group meeting (WP3). This meeting will be open to presentations from any project around CDK-based tools, both about the tools and about scientific research using those tools.

The funding will be used to fund the work of three researchers, Alyanne De Haan (orcid:0000-0003-0896-0906), Marc Teunis (orcid:0000-0002-3496-6669), and Egon Willighagen (orcid:0000-0001-7542-0286). Willighagen is co-founder of the Chemistry Development Kit and researcher at Maastricht University and has been involved in many other Open Science projects, involved in the Dutch Open Science Community, former editor of the Journal of Cheminformatics, and leader of various open scientific software projects. Teunis and De Haan have extensive knowledge about open educational resources, version control, and generally open science approaches to support research.

This project will focus on software, but if existing data is used, it will be open, to ensure compatibility with the open license of the CDK.

This is not expected, but if this is the case, it will be shared via Zenodo.

No. All the necessary resources (financial and time) to store and prepare data for sharing/preservation are or will be available at no extra cost.

The Chemistry Development Kit has a wide user base that uses the CDK directly or indirectly via one of the CDK-based tools. The Java package “org.openscience.cdk” is mentioned almost 50 thousand times on GitHub, the software using the CDK cited many times, and searching “chemistry development kit” on Google Scholar finds more than two thousand hits. The potential is very significant.

The CDK is released under the GNU Lesser GPL license, which is similar to the GPL license but allows use in proprietary software, and only changes to the CDK itself need to be made available under the same open license. Our work on the CDK library will use the same LGPL license as the CDK.

Similarly, our updates in WP1 will be made available under the same license as the tool it improves. An overview of licenses of these tools is given in Table 2.

Expertise is needed about version control, continuous building, code refactoring, software testing, software build systems, packaging standards, and other standards for software development. The applicants have this expertise, as demonstrated in their research output.

While it is not expertise needed, for formal releases of software, we will depend on the release managers of the respective packages. Some tools, like KNIME, have a fixed release schedule, but many do not. For the CDK itself, the release manager currently is Dr. John Mayfield. Deployments of CDK ‘snapshot’ releases to Maven Central are done by the main applicant. Of the tools we seek to improve, we know most developers personally (see co-authored articles in the publication lists of the applicants).