March 3, 2017

Commercialisation of scientific software; the case for open-source.

Since the Fort Lauderdale agreement in 2003, genomics has had perhaps the most 'open' attitude to data sharing of any field of science, and this is reflected in attitudes to software, where open-source is the norm.

 

 

Much like the Apache web server is invaluable to the internet, scientific open-source software, from aligners to genome browsers, are invaluable to bioinformatics. However, whilst some open-source software has proved valuable financially (e.g. the MySQL database), commercialisation of similarly-licensed academic projects has been almost non-existent. Here are three reasons why;

  1. Providing commercial support is way outside the scope of the academic groups writing the software.
  2. University tech transfer departments find open-source unfamiliar, and can see no way of monitising it, and
  3. Venture capitalists find the thought of investing in "open-source" companies somewhat alarming (for fairly obvious reasons).
There are exceptions (please convey others that I've missed), such as the KNIME workflow platform, and Revolution Analytics, who provide commercial support for the R statistics package. Revolution is particularly interesting as they managed to raise over $9M venture funding, and have a somewhat thorny relationship with some of the core R development team.

 

 

In my view, academic groups who want to commercialise their software should think seriously about entering into a collaboration agreement with an appropriate company, thus helping to;

  1. Increase their user base beyond the academic world,
  2. Demonstrate direct translation of research into industrial application (an important consideration for funding bodies),
  3. Increase software quality through collaborative development (i.e. external contributions of bug fixes and documentation) and,
  4. Support the group financially through revenue sharing.
Such agreements are, in our experiance, extremely powerful devices for getting academic software accepted by industry, and also for getting  us accepted by academia. I'm convinced that this approach is essential to enable some very brilliant scientific software reach its full potential.

Topics: Bioinformatics