Camelia, the Perl 6 bug

IRC log for #bioclipse, 2009-07-27

| Channels | #bioclipse index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
05:28 ChanServ joined #bioclipse
06:54 sneumann joined #bioclipse
06:58 miguelrojasch1 joined #bioclipse
07:07 mgerlich joined #bioclipse
07:58 masak joined #bioclipse
08:07 olass joined #bioclipse
08:29 egonw joined #bioclipse
08:34 egonw moin
08:34 zarah hello egonw, you fantastic person you
08:35 egonw zarah: you always point that out...
08:35 zarah egonw: you've just exceeded my capabilities :/
08:35 egonw makes me feel a bit akward....
08:35 egonw it's true of course... but still
08:41 olass hi egonw
08:41 egonw hej
08:41 masak hi egonw, you fantastic person you
08:41 olass egonw: I'd like to ask about the BODO
08:41 egonw masak: :)
08:41 egonw yes, please do
08:42 egonw won't be online much today either...
08:42 olass egonw: and unfortunately many things are broken in 2.0 release due to the recently discovered atom typing problems...
08:42 egonw family obligations...
08:42 olass oh
08:42 egonw but will hack-a-lot tomorrow
08:42 olass great!
08:42 olass ok
08:42 egonw yes, I will look at that top priority
08:42 olass about ontology
08:42 egonw yes
08:42 olass I am having problems wih BODO
08:43 olass BODO has bondCOunt
08:43 olass so
08:43 olass a parameter is needed to define bondCount order='s'
08:43 olass but that is not in the ontology
08:43 olass why?
08:44 egonw the order is not...
08:44 egonw yeah, I see your point
08:44 egonw it indeed should be...
08:44 olass what would make sense is that BODO has singleBondCount and CDK implemented that as bondCount with parameter 's'
08:44 olass or
08:44 olass BODO has parameters
08:44 egonw no, don't agree with that...
08:44 olass which one?
08:44 egonw the algorithm is bond count, and has parameter do tune the output
08:45 egonw the implementation is singleBondCount
08:45 egonw it is just impossible to make those 'subclasses' actual subclasses
08:45 egonw take the RDF descriptor
08:45 egonw radial distribution function
08:45 olass well the CDK implementation is implemented as bondCount with parameter 's' now...
08:45 egonw which was resolution and range as parameter
08:46 egonw in your above sketch, each combo would be a separate node in the ontology
08:46 egonw that simply does not scale
08:46 olass ok
08:46 olass what is the solution?
08:46 olass What we have now simply is not good enough IMO
08:46 olass add parameters to BODO?
08:46 egonw I need to formalize those parameters in the ontology
08:46 olass that sounds like quite some work...
08:47 edrin joined #bioclipse
08:47 egonw not necessarily
08:47 edrin hi
08:47 zarah hello edrin, you fantastic person you
08:47 olass egonw: I'd like us to settle an ID in the way I wrote in the email
08:47 olass http://www.blueobelisk.org/ontologies/chemoi​nformatics-algorithms/#bondCount&order='s'
08:47 zarah olass's link is also http://tinyurl.com/6n5gee
08:47 olass http://www.blueobelisk.org/ontologies/chemoi​nformatics-algorithms/#bondCount&order=s
08:47 zarah olass's link is also http://tinyurl.com/6n5gee
08:47 egonw well, that would not be an ID for the bondCount descriptor
08:48 egonw but could be an ID to the implementation
08:48 olass of course not
08:48 olass no
08:48 olass oh no
08:48 egonw though I am not sure the above will work int he RDF world
08:48 egonw (resource description framework, now)
08:48 olass it would uniquely define that I want to calculate number of single bonds
08:48 egonw moreover, the same problem arises
08:48 egonw combinatorial explosion of IDs
08:49 olass you will get that anyway, since you add parameters
08:49 egonw yes, but there is no need to collapse this into a URI
08:49 olass why not?
08:49 egonw because there are other solutions
08:49 olass please tell me
08:49 olass and tell me now
08:50 egonw I though I saw your QSAR ML already did this, not?
08:50 egonw guess not
08:50 olass no
08:50 olass well
08:50 olass sort of
08:50 olass but that is a custom solution
08:50 egonw so what?
08:50 olass How do we declarativiely say we calculated number of single bonds?
08:51 olass implementation-inspecific?
08:52 olass egonw?
08:52 egonw the current ad hoc solution, or using OWL, expressing the parameter types in the OWL
08:53 olass egonw: that is bad
08:53 olass a serious flaw in BODO
08:53 egonw which one?
08:53 olass :(
08:53 olass :(
08:53 olass :(
08:53 zarah :(
08:53 egonw so, we need to fix that
08:53 olass yes
08:53 olass and fast
08:54 olass if not, I will have to write my own
08:54 olass and would not be good
08:55 * olass is disappointed
08:56 egonw ha, that's what a PhD should be feeling...
08:56 egonw after all, he is supposed to touch the edge of science
08:56 olass I thought BODO was in better shape
08:56 olass now I have to remove all descriptors with parameters
08:56 olass until we sort his out
08:56 egonw better shape?
08:56 egonw the shape is excellent
08:56 egonw you are just asking new questions
08:56 olass no parameters?
08:57 olass new questions?
08:57 egonw it's just how you model the world
08:57 egonw realize that Dragon does not have parameters
08:57 egonw only the CDK did
08:57 olass reduce singleBondCount to bondCount + parameter but don't implement parameter?
08:57 egonw and perhaps JOELib
08:57 egonw it works fine for the latter two...
08:58 olass I thought the intention was not to just support CDK+Joelib...
08:58 egonw olass: please explain me the difference between a parameter defined in the OWL, or by the implementation
08:58 olass you want ME to do that?
08:58 olass you have skype?
08:58 egonw yes, because you feel you cannot do something which we have been doing in the past
08:58 olass you have skype?
08:59 egonw yes, 5 minutes
08:59 olass great
08:59 egonw as in ...
08:59 egonw in 5 minutes
08:59 olass :)
09:06 olass @time
09:53 miguelrojasch1 http://mspecs.tu-bs.de/
09:53 miguelrojasch1 maybe interesting for the bioclipse people
09:58 egonw_ joined #bioclipse
10:02 egonw_ how's the network here?
10:10 edrin left #bioclipse
10:38 ChanServ joined #bioclipse
11:28 masak egonw_: the good people at freenode reported some disturbances earlier today.
12:54 edrin joined #bioclipse
13:11 edrin zarah: ping
13:11 zarah edrin: pong
14:48 ChanServ joined #bioclipse
14:48 egonw_ joined #bioclipse
14:48 edrin joined #bioclipse
14:48 olass joined #bioclipse
14:48 masak joined #bioclipse
14:48 mgerlich joined #bioclipse
14:48 miguelrojasch1 joined #bioclipse
14:48 sneumann joined #bioclipse
14:48 CIA-63 joined #bioclipse
14:48 miguelrojasch joined #bioclipse
14:48 ilogger2 joined #bioclipse
14:48 stain joined #bioclipse
14:48 zarah joined #bioclipse
15:27 sneumann_ joined #bioclipse
16:01 egonw joined #bioclipse
16:27 egonw @tell masak http://plindenbaum.blogspot.com/2009/07/dr​awing-circular-genome-chapter-2-java.html
16:27 zarah egonw's link is also http://tinyurl.com/m925vx
16:27 zarah Consider it noted.
17:38 olass joined #bioclipse
17:38 olass moin
17:38 zarah privet olass
18:08 sneumann_ joined #bioclipse
18:28 edrin left #bioclipse
18:37 sneumann__ joined #bioclipse
18:51 wins joined #bioclipse
19:42 sneumann joined #bioclipse
20:13 ChanServ joined #bioclipse
20:41 egonw joined #bioclipse
20:58 sneumann joined #bioclipse
21:21 olass @pz 1439
21:21 zarah http://pele.farmbio.uu.se/cgi-bi​n/bugzilla/show_bug.cgi?id=1439
21:49 CIA-63 bioclipse.cheminformatics: Ola Spjuth master * r2613862 / (2 files): Made perceiveSybylAtomtypes return String, just like perceiveCDKAtomTypes and in accordance to its javadoc and manager help. - http://bit.ly/10qAUh
21:49 CIA-63 bioclipse.cheminformatics: Ola Spjuth master * rc8c89c4 / (2 files in 2 dirs): Updated tests for changed API in perceiveSybylAtomTypes(). - http://bit.ly/bY2eG
21:50 egonw olass: your patches are applied to master
21:50 olass thx
22:31 egonw OK, I isolated the problem with the atom typing in the CML file
22:31 egonw I believe this is a genuine CDK bug
22:45 olass egonw: oh
22:45 olass glad you isolated it!

| Channels | #bioclipse index | Today | | Search | Google Search | Plain-Text | summary