[octet-devel] Chemical Semantic Web and Thanks

[Peter Murray-R. <pm...@ca...>]

[Crossposted to several OpenSource lists - please reply thoughtfully. I 
will collate significant replies]

I presented an invited lecture at the Royal Soc Chem yesterday on "The 
Chemical Semantic Web" and highlighted the impressive contribution of the 
Open Source movement in chemistry. I hope I managed to acknowledge most 
people's contributions. It generated a lively discussion in which, I think, 
the Open Source approach came out very positively.

Two common negative questions were presented:

"Who pays for it"
  the implication was that if there was no commercial development the 
software was second-class and there was no future for it. My response was 
that the code would be eagerly adopted by the bioinformatics community 
(strongly confirmed by Mike Ashburner) and we are starting to set up strong 
links with the Bioinformatics community on small molecules. This question 
came from the pharma industry and so I challenged them to put something 
back into the communal process. It's noticeable that the pharma industry is 
a consumer of our efforts but has not contributed anything in return and 
continues to pay for overpriced software products that do not use modern 
methodology. (I spent 15 years in Glaxo so I feel I have a right to 
criticise the industry). I know that pharma industry reads this list and I 
would be delighted if there were some who wished to support communal 
activity. [I would suggest they mail me directly and I will summarise 
replies. All replies will be assumed to be Open unless specifically 
indicated]. One approach would be to support a physical workshop on 
chemical open source. Particular topics could be:
- markup languages
- communal chemical ontologies (e.g. descriptors)
- Open datasets
- use of InChI (we are very excited about this)
- graphical components
- common code libraries
- web services (I'll mail later)
The only contribution that I know of from the pharma industry to 
chemoinformatics is the JME editor which is free, but despite repeated 
requests is not Open and not likely to be. It could be withdrawn at any time.

"Open source is unfair to commercial developers"
I think this attitude is particularly strong in the US and perhaps Germany. 
The idea is that public funding should not be used to compete against 
commercial organisations. As Mike Ashburner replied, most Open software in 
bioinformatics is written by people on short term contracts often with no 
possibility of extension. I know that made contributors to these lists do 
not have jobs. I developed much of CML when I was unemployed after being 
made redundant from Glaxo.
I would be more (slightly) more sympathetic to this view if the chemical 
software industry provided anything substantial to the community. I know of 
no support for Open efforts from the chemical software industry other than 
a very undersupported effort to get an Open Management Group spec for 
chemistry. I was invited to contribute CML as the core of this spec, but 
even then it was made clear that I would not even have my (foreign) travel 
expenses paid - even though I had no job.
Note that "free" is not good enough and may even by counterproductive. 
Closed software often stifles development as there is no API and the 
software can be withdrawn at any time. That is why we are so keen to see a 
JChempaint component as a universal Open chemical editor for the web.

In fact I think the tide is turning. The components that you are all 
developing are ideal for Web Services, which will have the potential to 
rapidly become a major player in the chemoinformatics market. Between the 
members of this community we have standards, data, transformers, 
algorithms, repositories, query languages, graphics and much more. I'll put 
forward a strategy in the next mail(s).

Meanwhile - when you are hacking the latest set of bugs at 0200 in the 
morning, take heart - it really is appreciated - and yes - the day of Open 
Source for "chemistry" is dawning. In fact I think the centre of gravity of 
chemical software development is starting to shift towards bioscience [and 
materials] so that people will simply refer to the informatics of small 
molecules as "bioinformatics"



Best

P.

Many thanks to those who have sponsored our Open Source development: Royal 
Soc. Chemistry, Nature Publishing Group, Cyprotex. More about this later.



Peter Murray-Rust
Unilever Centre for Molecular Informatics
Chemistry Department, Cambridge University
Lensfield Road, CAMBRIDGE, CB2 1EW, UK
Tel: +44-1223-763069