Question: Analyse Microarray Data - AVG_Signal
0
gravatar for Alex Greenshields-Watson
4 weeks ago by
Cardiff

Hello,

I have a historic microarray dataset from an illumina bead based technology. For my data I only have an dataframe where the columns include, for each sample:

"AVG_Signal"
"NARRAYS"
"ARRAY_STDEV"
"Detection Pval"

I would like to run a comparison analysis using limma to get p adjusted values, however I am not sure what my data corresponds to based on instructions in the limma users guide.

I am more familiar with RNA-seq analysis using DESeq2. However have read posts which suggest Deseq2 is not applicable for microarray - is this correct?

What should I do regarding normalisation and comparison?

The within array normalisation function does not seem applicable.

I have proceeded straight to model.matrix(), lmFit() and makeContrasts() - without any normalisation - however I have only a very small number of differentially expressed genes as a result.

Any advice would be very helpful.

Kind Regards, Alex

microarray limma • 101 views
ADD COMMENTlink modified 4 weeks ago by Gordon Smyth39k • written 4 weeks ago by Alex Greenshields-Watson0
1

Correct. DESeq2 is not appropriate for microarray.

ADD REPLYlink written 4 weeks ago by Michael Love26k
Answer: Analyse Microarray Data - AVG_Signal
2
gravatar for Gordon Smyth
4 weeks ago by
Gordon Smyth39k
Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia
Gordon Smyth39k wrote:

The limma pipeline for Illumina Beadchips is described in Sections 4.6 and 17.3 of the limma User's Guide. Briefly, you use read.ilmn to read the file and neqc to background correct and normalize.

limma will read the intensities (AVE_Signal) and the detection p-values, and will use the detection p-values to infer the intensities of the background probes as part of the background correction step. Usually the NARRAYS entries will be all 1.

ADD COMMENTlink modified 4 weeks ago • written 4 weeks ago by Gordon Smyth39k

Many thanks. Apologies I missed this in the guide ;) Kind Regards, Alex

ADD REPLYlink written 4 weeks ago by Alex Greenshields-Watson0

I know this is a slightly separate question - but if I want to then use this normalised Limma data in visualisation, PCA or distance analysis - should you sum or average multiple probe values that correspond to the same gene? Or just leave them as they are?

ADD REPLYlink written 4 weeks ago by Alex Greenshields-Watson0

Found this post:

https://support.bioconductor.org/p/43745/

Sorted thanks.

ADD REPLYlink written 4 weeks ago by Alex Greenshields-Watson0
Answer: Analyse Mircoarray Data - AVG_Signal
0
gravatar for James W. MacDonald
4 weeks ago by
United States
James W. MacDonald52k wrote:

If you just have averages, you can't do much with those data. You need the raw data.

ADD COMMENTlink written 4 weeks ago by James W. MacDonald52k

Well, I think it's technically the average signal per bead for each gene in that sample. From looking online this seems like a standard output from illumina bead arrays and I think this is in fact the raw data (as a non-normalised bead summary), in a format created by bead studio, see publication below - raw data file.

http://www.damtp.cam.ac.uk/user/st321/CV&Publications_files/STpapers-pdf/DSTT06.pdf

As a follow up I've managed to get it through Limma and have now got a meaningful set of DEGs that make sense biologically. I will check and then post the code in a follow up.

ADD REPLYlink written 4 weeks ago by Alex Greenshields-Watson0
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 139 users visited in the last hour