expression correlation on an ExpressionSet
2
0
Entering edit mode
DataFanatic ▴ 10
@datafanatic-12212
Last seen 2.2 years ago

Hi, I have .txt files with phenotype data and expression values from an Affymetrix microarray. I constructed an ExpressionSet class from scratch. I need to generate and plot an expression correlation matrix for all genes comparing case vs control.I'm very new to R and Bioconductor and would greatly appreciate as much detail in your answer. Thanks!

ExpressionSet caseControl Correlation Plot • 1.1k views
0
Entering edit mode

Detail in your question would be appreciated as well. What exactly is an expression correlation matrix?

0
Entering edit mode
DataFanatic ▴ 10
@datafanatic-12212
Last seen 2.2 years ago

I want to know if the expression of all genes in the controls differs from those in the cases.  I would like to visualize this with a heat map.

0
Entering edit mode

To answer your question, see the vignettes for ComplexHeatmap. But do note that a heatmap isn't inferential. In other words, just because something looks different on a heatmap doesn't imply that it really is different. In general it is a better idea to use something like limma to determine which genes show evidence for being different between your two groups, and then use the set of differentially expressed genes to make the heatmap.

An example of what I am talking about is this paper. It's RNA-Seq, and uses heatmap.2 from gplots, but the idea is the same.

0
Entering edit mode

Thanks!  I was told that to do an expression correlation (case/control) all I needed was Excel. Can you give me your thoughts about my approach, I started with .txt files and build an ExpressionSet ( is this what people normally do?) I did this because reading the Affy and other vignettes I was able to understand that data is usually formatted as an ExpressionSet. I am also using geNetClassifier to so a more comprehensive analysis of the ExpressionSet.

0
Entering edit mode

Now that's hilarious. Of course you only need Excel. That's all anybody needs.

Ideally you would have the celfiles and would use oligo or xps to read in and summarize the data. Then you could use limma to make comparisons.

But you seem to be pretty lost, so you are going to need to either find somebody local who can help you with this analysis, or you are going to have to buckle down and start reading. There are vignettes for the oligo and limma packages (the limma User's Guide is particularly comprehensive), plus several papers in the Bioconductor F1000 channel (I gave you a link already).

What you won't be able to do is ask enough questions on this support site to get you through the analysis. Analyzing data is just too complicated for that, plus people will quickly get bored and stop wanting to help you.

0
Entering edit mode
vulegom • 0
@vulegom-13057
Last seen 5.7 years ago

happy birthday wishes