CCL: Another Pipeline Pilot Question: Loops
- From: "Van Hoorn, Willem"
<Willem.Van.Hoorn],[pfizer.com>
- Subject: CCL: Another Pipeline Pilot Question: Loops
- Date: Fri, 8 Feb 2008 13:34:57 -0000
Sent to CCL by: "Van Hoorn, Willem" [Willem.Van.Hoorn[*]pfizer.com]
I presume you use the "Learn Good Molecules" component (Bayesian
learning)? By default the 'Validate Models' option is turned on. This is
a leave-one-out cross-validation, the results of which will be in the
help text of the model component (which will be in your
LearnedProperties tab if your model building was successful).
In general, if you want a loop in Pipeline Pilot you will have to put
whatever you want to be inside the loop in a subprotocol for which you
have set the RunToCompletion to True.
There is a download library on the Scitegic/Accelrys website:
http://accelrys.org/pipelinepilot/index.html
________________________________
> From: owner-chemistry{}ccl.net [mailto:owner-chemistry{}ccl.net]
Sent: 07 February 2008 03:23
To: Van Hoorn, Willem
Subject: CCL: Another Pipeline Pilot Question: Loops
Hi all,
I was wondering if any one would be able to help me. I can not figure
out how to "loop" over a PipeLine pilot protocol that I have written.
I
would like to run it a number of times and generate some statistics.
I have written a protocol to build a model to classify drugs from
non-drugs.
It takes in actives/inactives and splits them into a training and test
set. The model is then build using the training set, and then validated
on the test set.
I am wondering if there is any way I can loop over this procedure, to
turn it into a cross validation procedure?
As an aside, does anybody know of a good resource for example PipeLine
pilot protocols? This is all rather new to me!
Thanks for any help :)
Iain