History

The original NMRShiftDB project started after some initial attempts by Christoph Steinbeck in 2002, when Stefan Kuhn was hired to develop it at the Max-Planck-Insitut für chemische Ökologie in Jena. MPI-CE also provided hosting for some time. NMRShiftDB 1.0 was available from November 2003, running on a failsave cluster at MPI-CE (this setup was described in the first NMRShiftDB paper). At the end of 2003, there were more than 6000 searchable spectra in the database, 10,000 spectra were available in mid-2004. The Steinbeck group had moved to Cologne University Bioinformatics Center (CUBIC) in 2003.

During this time, the philosophy of a peer network of servers was developed, which should help spreading usage by groups having their "own" server. Therefore, a second node was set up at CUBIC and for some time Marburg University also operated a node. In practice, however, the idea did not really work and caused more work and problems than real benefits.

In 2005, NMRShiftDB had been one of the first contributors to the newly established PubChem database, where the data have been deposited ever since. In 2007 a co-operation with the Chemspider group started. This co-operation also lead to more work on the qualitiy of the data. We also got in contact with Wolfgang Robien leading to some fruitful, albeit sometimes heated, discussions.

Another attempt during this time was to establish a standalone client, enabling a better user interface and offline data collection. Unfortunately, both failed. Firstly, we tried a Swing-based application, but judging from the non-existent feedback, that was not a success. Later Speclipe, a collection of plugins for Bioclipse project (Bioclipse was co-developed by the Steinbeck group and uses CDK a lot), was started. Also here, feedback was low and requests from Speclipse to the servers were sparse. Whilst Speclipse is still available for download, it is no longer developed AFAIK (I=Stefan Kuhn) and I don't think it will. Main reasons are that due to the Bioclipse project organisation it is virtually impossible for external people to develop for Bioclipse and I also believe that Bioclipse is effectively dead (a shame, it was a nice project once).

During the time at CUBIC we also started the lab administration component of NMRShiftDB. It was developed in close co-operation with the NMR lab of the Chemistry Department at Cologne University and the lab has been running all its user interactions via it since 2007.

After CUBIC ceased to operate, in 2008 the CUBIC server was moved to NMR labs of the University of Mainz. In 2009 Christoph Steinbeck and Stefan Kuhn started new positions at European Bioinformatics Institute at Hinxton, UK, eventually installing NMRShiftDB in the EBI data centre. The server in Mainz was retired at that point.

In 2010 Stefan Kuhn left EBI and Christoph Steinbeck decided not to support the project any more. Following this, some interested people got together and continued the project under the name nmrshiftdb2. Initial hosting was done in Mainz again, in 2011 the NMR lab in Cologne provided a new server. Since then, the software has been maintained and slightly extended (e. g. inclusion of original spectra), the database has been extended and maintained. We also continued development of the lab system. In 2012, we are close to 41,000 structures and almost 50,000 spectra.

In November 2013 we could celebrate 10 years of NMRShiftDB/nmrshiftdb2, with 42,000 structures and 51,000 spectra.

A noteable feature added was the stereo-enhanced prediction using a stereo-enhanced HOSE code, which got published in 10.1021/acsomega.9b00488. There were also several papers using nmrshiftdb2 for machine learning, in particular deep learning approaches once those became available. Work in the field of data formats led to the development of NMReDATA, where nmrshiftdb2 is a collaborator. ID NMR was also heavily inspired by nmrshiftdb2.

Hosting continued in Cologne. In 2023, approaching 20 years of nmrshiftdb2, there were about 260,000 structures and 450,000 spectra. The papers about nmrshiftdb2 collected 372 citations so far in google scholar. For the 20th anniversary, we managed a paper in MRC.

So nmrshiftdb2 is an evolving project. What will the future bring? That obviously depends on the time and material we can put into it. For big ideas we would need extra funding, but suggestions and co-operations are always welcome. And of course, anybody interested in participation is welcome as well. It could be your project as well.