[OTDev] Batch processing with ToxPredict | Fwd: Uncaught bounce notification

Vedrin Jeliazkov vedrin.jeliazkov at gmail.com
Fri Apr 27 11:20:38 CEST 2012


Hi Sylvia, All,

On 27 April 2012 10:27, Micha Rautenberg <mr at mrautenberg.de> wrote:
> I resend the misrouted mail from Sylvia, Micha
>
> Dear Nina, dear Christoph, dear all,
>
> I tried to upload a file of structures (in different formats, as indicated
> in the legend of the tool) to Toxpredict for batch processing (initially
> about 100 structures, then 10 for testing). Although I was logged in, it
> did not work. Single structure input worked well.
>
> Di you have any advice for me, for which reasons the batch processing was
> not possible?

I uploaded a dataset (sdf file, attached), consisting of 10 records.
After the upload was finished, ToxPredict displayed a page with
details on the first compound in the dataset. I then clicked "Browse
all" (located on the left hand side of the screen) and ToxPredict
displayed another (tabular) view of the dataset. This is the place
from which one can launch batch processing of datasets by pressing the
button "Run models" at the top of the screen. I proceeded with this
step and after a while obtained the results in the same tabular view
of the dataset (you have to select the "show predictions" option in
order to view them). Of course, you can also download the resulting
dataset that includes these model predictions.

We do agree that the user interface is not sufficiently intuitive and
could be improved, but the batch-processing functionality is there and
it is working fine.

Kind regards,
Vedrin

PS: While performing the above tests, I've noticed that some IBMC,
NTUA and TUM models are either returning errors (IBMC) or not
accessible (NTUA, TUM), which has the nasty side effect of slowing
down ToxPredict. It would be nice if those problems could be fixed,
otherwise we might have to disable querying those models from
ToxPredict. Here are some more details on the models affected:

-- IBMC:

EPAFM LC50 fish OpenTox model created with SCR regression model
http://195.178.207.160/opentox/model/makeSCR_457
http://195.178.207.160/OpenTox/algorithm/MakeMNA unknown protocol: asdasdasdhttp

-- NTUA:

MLR model for caco2
http://opentox.ntua.gr:8080/model/059d8bf0-cccc-4ad1-ad40-d15a663894fc
[500] Connection refused

MLR model for Exp LogKow
http://opentox.ntua.gr:8080/model/192f9d28-cb99-4837-aad6-ba4f348adee9
[500] Connection refused

-- TUM:

OpenTox model created with TUM's J48 model learning web service for
Micronucleus Data.
http://opentox-dev.informatik.tu-muenchen.de:8080/OpenTox-dev/model/TUMOpenToxModel_j48_3
[500] Connection refused

OpenTox model created with TUM's kNNregression model learning web
service for the Caco-2 Permeation endpoint
http://opentox-dev.informatik.tu-muenchen.de:8080/OpenTox-dev/model/TUMOpenToxModel_kNN_10
[500] Connection refused

OpenTox model created with TUM's kNNregression model learning web
service for carcinogenicity
http://opentox-dev.informatik.tu-muenchen.de:8080/OpenTox-dev/model/TUMOpenToxModel_kNN_13
[500] Connection refused
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test.sdf
Type: application/octet-stream
Size: 26233 bytes
Desc: not available
URL: <http://lists.opentox.org/pipermail/development/attachments/20120427/8e8e9638/attachment.obj>


More information about the Development mailing list