[Bio-software] Re: GCG non-support

Peter Rice via bio-soft%40net.bio.net (by pmr from ebi.ac.uk)
Fri Aug 3 04:18:27 EST 2007


Steve Thompson wrote:
> Hello Peter - Thanks for joining the dialogue.  I was hoping you would.
> 
> Yes, I have heard reports of extensive SeqLab customizing, and in fact, 
> Lynn Miller at GCG tried to spearhead a SeqLab extensions public 
> library, but she was never able to get folk to contribute.  Alas, the 
> project shriveled away.  It seems as if most of this work was done 
> behind the closed doors of the corporate world where sharing is not the 
> norm.

In this case, sharing may be possible. I can ask anyway.

> Yes, Eric Linton, MacGDE's maintainer, was a student here at the Marine 
> Biological Laboratory's Workshop on Molecular Evolution, where I am 
> working this month, many, many years ago when he began thinking about 
> this MacGDE project.  Another of our past Mol Evol Workshop students, 
> Tulio de Oliveira, has created a robust Linux port (See 
> http://www.bioafrica.net/GDElinux/index.html).  Both a wonderful pieces 
> of work, but neither includes all of the SeqLab's functionality.

Can you make a quick list of the additional functionality you would like 
to see?

I seem to recall "rich sequence format" was one of GCG's major extensions.

> It describes some vague "perpetual license" concept that leaves many 
> questions unanswered.  It does not offer source code.  It does not offer 
> any pricing information.  And, perhaps most importantly, it offers no 
> method of transferring the current license, which has been tied to a 
> specific node that the package is installed on ever since Accelrys took 
> over, to a new node, when that old node becomes obsolete or breaks.  
> I'll post it on my Web system: http://bio.fsu.edu/~stevet/GCG.death.pdf.

Interesting. I see it mentions Pipeline Pilot. EMBOSS is an ISV partner 
and committed to interfacing EMBOSS applications as Pipeline Pilot 
components.

I hope there is some source code release. We had the source code at 
Sanger not just for EGCG - but also because the sequencers needed more 
than the 350kb limit on sequence length.

But I see no quick way to decide on possible source code licensing. Too 
many authors, too many companies/institutes. I am not surprised that 
nothing is promised at this stage.

Hang on to your licences though ... one (legally speaking) relatively 
easy possibility would be a cheap source license for existing licensees.

> I again agree - LookUp has always been problematic.  SRS 5 is way 
> better, especially if it could be integrated into SeqLab's list file 
> mechanism.

That would be difficult ... and I am not too confident of the status of 
SRS5 code anyway. But there are some alternatives around.

>> When EMBOSS started I had to kill EGCG. GCG claimed rights to the code 
>> and refused to allow its reuse, so I claimed rights to the code and 
>> refused to allow them to continue distribution. All the useful 
>> applications had to be rewritten from scratch for EMBOSS. Only 
>> prettybox survived - Rick Westerman contributed it to GCG.
> 
> Yes, we all understood this regrettable but necessary maneuver.  This 
> was right about when Oxford Molecular took over GCG, wasn't it?  And 
> coincident when they made the decision to no longer distribute source?

Well, there was at least a year when I tried to make all the code 
changes needed for 9.0 and worked closely enough with GCG for the EGCG 
users to upgrade to version 9.0 (many were using the external users 
licence and had no real choice).

The problems were with the code itself. It took a lot of effort to 
convert to 9.0, which made it worth the effort to write new libraries 
and even rewrite applications for EMBOSS - 3 years for 2 developers to 
reimplement EGCG.

Source code licensing was a problem too ... I had a special licence that 
allowed me to distribute code to other sites who had source code ... but 
they would still be unable to send code to me.

Ideally I would have given EGCG to GCG in exchange for the right to 
reuse my own code, but that never happened. A pity, as it cost us no 
extra effort in the end to reimplement and a lot of GCG users lost out.

Oxford Molecular were as helpful as they could be at the time, but all 
the decisions were left to GCG. In those days open source was something 
that scared companies, so I was not surprised by the result. Sad though, 
as it immediately made EMBOSS into a competitor rather than a partner. 
It has taken us 10 years (and some luck) to become an Accelrys partner.

>> There was a release 9.0 of EGCG (despite the GCG source code licence 
>> problems) but it only ever went for beta test at one institute, and 
>> that institute is now closed down.
> 
> Yes!  When I was at Washington State University's VADMS Center with 
> Susan Johns, she contributed to EGCG and I think I remember testing 
> v.9.0. But, you're correct, VADMS is no longer an entity (admin' never 
> fully funded nor appreciated it - seems we were ahead of our times).

Hah! Not the one I meant (VADMS had an alpha binary release). The beta 
test was at HGMP/RFCGR in Hinxton - where the EMBOSS development team 
was and it was their closure that threw EMBOSS into crisis mode.

Susan's TOPO was one of our first EMBASSY package add-ons, along with 
Will Gilbert's MSE editor.

An early version of MSE was used by GCG for several of their editing 
applications, so at least that and GDE 2.3 are currently available with 
source code. EMBOSS's MSE is under GPL. I will try to figure out what 
GDE's licence really means.

Peter Rice
European Bioinformatics Institute



More information about the Bio-soft mailing list