Lda To Search Biomarkers

Question

[Elisa]

Dear all

I'd like to perform some statistical analysis on a 2d gels to find biomarkers. I've read that many authors apply Linear Discriminant analysis. This is a statistical classification tecnique based on Bayesian classification. I have difficulty to understand how to apply it on gels to find biomarkers.

Could you give me a simple example?

Bests

Answer 1

[Neilfws]

Suggest you start with this review (PDF): "Multivariate Statistical Tools for the Evaluation of Proteomic 2D-maps: Recent Achievements and Applications." It's not a great article, but covers LDA and points to some more useful references.

For understanding LDA itself, a good starting point is the Wikipedia entry.

LDA is a fairly simple idea. You have two or more classes - let's say "normal" and "cancer". Your aim is to identify a linear combination of variables which best characterizes the classes.

So the question is: in the case of 2D gels, what are the variables? The simplest example would be: the intensity of the spots. That's the approach used in this article: "Chemometrics of differentially expressed proteins from colorectal cancer patients", which employs both PCA and LDA.