[BioC] identification of disease groups
Saroj K Mohapatra
saroj at vt.edu
Thu Jun 11 16:49:26 CEST 2009
> ... where we have 5 normal samples (untreated)
> with 6 patients treated with drug A and 5 patients treated with drug A +
> B. The biologists question is there any serious perturbations of the genes
> between two groups (Treated with A only and treated with A+B).
I agree with your choice: by applying a p.value/ratio threshold in limma
one can find the genes affected by drug A (A-Control) and genes affected
by drug A+B (AB-Control). Union of the two sets would have all the genes
differentially expressed (d.e.) by EITHER treatment; intersection would
have the genes d.e. in BOTH.
> They are
> interested to represent the same by a cluster.
Do you mean a heatmap from hierarchical clustering? You could select the
genes and then get the expression matrix for all three groups of
samples, order them by genes only (not samples) and the resulting
heatmap would have, hopefully, regions of gene expression showing
difference between control and treatments.
> I am just wondering, whether I should do separate analysis of those two
> groups using limma and cluster together by selecting genes using some
> parameters such as P value or fold value.?
> Whether I should do analysis together using limma and cluster it by
> selecting genes by fold value of P value.?
> It will be of great help if any one of you could provide suggestions on the
> Thanks in advance.
> [[alternative HTML version deleted]]
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
More information about the Bioconductor