Using PLS for diagnostic ?

Moderator: MultiD Support

Using PLS for diagnostic ?

Postby mrklomp » Tue May 29, 2012 4:49 pm

Hello everybody,

I didn't know where to post my question because it is a more related in general qPCR assay so Itry here.
I've yet to master all the subtetlies of the qPCR assays analysis, but I'm trying to create a mathematical classification model to discriminate samples in two conditions.

I've tried some things : I'm doing 2 parallel analysis :
1st one using the 2-ddCt Method.
2nd one using a PLS on the individual dCt obtained after using qpcR package.

Then using cut-off criterion I established a discriminant template, but I'm not quite sure what to do after. Maybe I should use another method instead of PLS analysis.

Actually all the information about qPCR diagnostic kit information is quite the taboo because of the commercial issue from the companies whom own these kits and they stay very evasive on how to interpret the results.

So i was wondering if you could share your insights regarding this issue.

Kind regards,

Mrklomp
mrklomp
 
Posts: 3
Joined: Mon Jan 02, 2012 11:50 am

Re: Using PLS for diagnostic ?

Postby Mikael Kubista » Tue Jun 05, 2012 2:59 pm

Its not obvious to me how you use PLS. We occasionally use PLS as implemented in GenEx and it can be used on either Cq,deltaCq or even deltadeltaCq values. It is not replacing pre-processing; PLS is standard method for multivariate calibration; essentially, you construct a single standard curve based on the response of many genes (which may be normalized to reference genes and may be normalized to reference sample).

Good luck
Mikael Kubista
 
Posts: 455
Joined: Tue Jul 01, 2008 12:28 pm

Re: Using PLS for diagnostic ?

Postby mrklomp » Wed Jun 06, 2012 9:02 am

Thank you for your answer Pr Kubista. I will use a more conventional clustering approach I think in order to propose a valid template of discrimination.
mrklomp
 
Posts: 3
Joined: Mon Jan 02, 2012 11:50 am

Re: Using PLS for diagnostic ?

Postby Mikael Kubista » Thu Jun 07, 2012 6:42 pm

It depends on objective. There are many powerful classification/clustering methods including PCA, hierarchical clustering, artificial neural networks, support vector machines etc., while PLS is the most powerful calibration method we have. For classification dynamic PCA will be available in next release of GenEx. We have had access to it for some time and it is very powerful.
Mikael Kubista
 
Posts: 455
Joined: Tue Jul 01, 2008 12:28 pm


Return to Partial Least Squares (PLS)

Who is online

Users browsing this forum: No registered users and 1 guest

cron



MultiD Analyses

Home of the GenEx analysis software




Partners






































www.Gene-Quantification.info