Exciting new software for electrophysiology – NeuroDebian@G-Node

Upon an invitation by Thomas Wachtler, on Feb 28 2012 NeuroDebian visited the German INCF node – G-Node for a short workshop on Debian packaging. G-Node drives a number of software projects primarily focused on electrophysiology research – ranging from server-side data storage to data access and analysis tools for workstations. None of those are part of (Neuro)Debian yet, so all participants had high hopes to change this as soon as possible.

During the workshop we looked into various projects and discussed how they could be packaged for Debian. Fortunately, it turned out that the majority of the necessary packaging work will be fairly straightforward and there are no obvious show stoppers, such as licensing problems, that would prevent inclusion into Debian in a rather near future. This is especially exciting as no less than three key infrastructure problems of electrophysiology research, and electrophysiology on Linux machines in particular will be tackled with the new packages.

Access proprietary data formats

Maybe one of the most pressing issues is data access. There are numerous hardware vendors for data acquisition devices that all have their own proprietary data formats. This is a serious obstacle when it comes to data sharing efforts or collaborative work in general. To access such data, one depends on the vendor to provide support for a particular computing environment. However, not surprisingly this support is typically limited to the Windows platform, or other commercial systems, such as Matlab.

For many years these hardware vendors could not agree on a common data format, but instead came up with a standardized API to access vendor specific IO libraries to ease the life of developers trying to add support for a data format to their software. This standardization initiative is called Neuroshare. Unfortunately, it does not address the core of the problem, as it still requires the vendors-specific libraries implementing the Neuroshare-API to access a proprietary data format – and of course the vast majority of these libraries is closed-source and Windows-only.

G-Node has developed a Wine_-base solution that behaves like a Neuroshare library on the Linux platform and internally passes library calls on to one of the vendor libraries. This nswineproxy can be used as a drop-in replacement for any Neuroshare library and therefore allows for access to all Neuroshare-compliant modules. While this solution admittedly lacks the performance of a native implementation, it still makes a huge difference to be able to access a particular data format at all, and is perfectly suitable for one-time data conversion – for example into the open HDF5 data format.

In addition to the Wine-proxy, G-Node also provides python-neuroshare, a Python package to access any Neuroshare-library.

Comprehensive data description and sharing

Plain data access is not sufficient for collaborative work and data sharing. In addition, it requires a comprehensive description of data, including information on data origin, acquisition parameters, and applied processing. G-Node offers several software packages that offer important functionality in this regard. ODML is a markup language to create comprehensive meta information for a particular dataset (Python and Java bindings are provided). datajongleur is a Python package to create data views that combine plain data with meta information, such as physical quantities, or more detailed meta data. It allows injecting these enriched datasets in SQL databases for efficient storage, querying, and retrieval. Finally, G-Node also develops a server-side database solution that aids multi-site data-sharing efforts. Clients for various environments, such as Matlab and Python are either being developed or are already available.

Real-time data analysis

The third topic for software development at G-Node are closed-loop real-time systems for online stimulus creation, data acquisition, and data analysis, such as Relacs. The upcoming Debian release (wheezy) will offer a Linux kernel with real-time capabilities which should make it easier to offer out-of-the-box solution for this type of research.

If you are interested in any of these software packages and would like to see them integrated into the Debian system quickly, please do not hesitate to contact us, so we can coordinate the effort.

Comments

blog comments powered by Disqus