[Owl-devel] Pisa connection

Jose M. Duarte jose.m.duarte at gmail.com
Mon May 31 13:07:46 EDT 2010

Hi Henning

Ok so I'll try to reply all of them:

On 31 May 2010 15:52, Henning Stehr <stehr at molgen.mpg.de> wrote:

> Hi Jose,
> the stuff in owl.connections.pisa looks amazing. Just some quick
> questions. What I'm interested in is getting the residues which are
> involved in protein-protein interfaces in the most likely assembly.
Right now I've only implemented the parsing of interfaces description and
not that of assemblies. It shouldn't be too difficult to parse the XML
assembly description files though.

> 1. Are the interfaces returned by getInterfacesDescription()
> automatically taken from the "best" assembly or all assemblies or some
> other default one?

Those are all possible interfaces found by the PISA algorithm

> 2. I assume that PisaInterface.firstMolecule and
> PisaInterface.secondMolecule are the two molecules which are in
> contact via the interface. If the assembly is e.g. a homo-dimer, would
> firstMolecule and secondMolecule then be the same?

Not necessarily. The thing is the PisaMolecule objects contain the whole
description of the interface including all the BSAs of residues involved.
Even in the case of a homo-dimer, the interfaces on both sides can be

> 3. Say I list the residues by PisaInterface.getFirst().getResidues().
> Are these only the ones in contact in this interface (and this
> assembly) or all residues in the chain/molecule?

They are all residues in the interface. To get the subset of contacting
residues you will have to filter by using the BSA values
(PisaResidue.getBsa()). Basically the getRimAndCore() method is doing that
but returning a PisaRimCore object which divides the contacting residues in
core ones (above the BSA cut-off) and rim (BSA>0 but below the cut-off).

4. Currently you are downloading pre-calculated data for existing PDB
> codes, right? How different/difficult would it be to get data for
> custom uploaded structures?

Right now all I'm doing is accessing the pre-calculated PISA data available
for download in XML format (see
http://www.ebi.ac.uk/msd-srv/prot_int/pi_download.html). Doing it for custom
structures would be quite difficult. It would require parsing the html,
dealing with cookies, well you know the whole pain of reverse-engineering
the site...

> 5. Maybe a question for the other Pisa experts on the list: Would it
> even make sense to run such delicate calculations as Pisa does on
> predicted structures?

I would say it doesn't make sense. Pisa needs pretty accurate interface
descriptions in order to work properly. But that's just a gut feeling, don't
quote me on that ;-)

> BTW, did you know there is also a standalone version of PISA in CCP4?
> (see http://smb.slac.stanford.edu/facilities/software/ccp4/html/pisa.html)

Yes, I tried that one also but then I realise it was very out of date and
doesn't seem to be getting updated anymore. I don't remember exactly but
there were something like 6 releases between that and the live one on the
PISA web site (with many important features added in between).

Hope it helps

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.bioinformatics.org/pipermail/owl-devel/attachments/20100531/8ee15de4/attachment.html>

More information about the Owl-devel mailing list