Proc Mixed - Differentially Expressed Genes

Question

[Priyanka Surana]

I am new to microarray data analysis. I am analyzing microarray experiment using PROC MIXED in SAS and use genotype, replication, time point, genotypetimepoint as fixed factors and replicationtimepoint as random factor. I want to find differentially expressed genes for each genotype*timepoint combination. Do I look at lsmeans file and associated p-values or diffs file and associated p-values.

Thanks, Priyanka

[Chris Evelo]

You might want to consider not to use SAS... It is of course possible to do microarray statistics in SAS but the mainstream approach really is to do it in R/Bioconductor and many complete packages exist to assist you in doing so. If you are new to the field, like you say you are, I would just switch to Bioconductor and learn that, even if you have prior knowledge of SAS.

[David Quigley]

Since your question is essentially about how to interpret a generalized mixed linear model and is not inherently tied to microarray data, you might consider asking it on stats.stackexchange.com, where the readership is mostly statisticians.

Answer 1

[Pablo]

In chip analysis, R is by far the "the facto standard". I would recommend switching to R since everything is already solved there (and this is from someone that really dislikes R).

There is a brief tutorial here (chapter 6 talks about micro-arrays).

Answer 2

[Cheng Zhongshan]

Maybe you can try Anovarray. AnovArray is a set of SAS macros for the Analysis of Variance of Gene expression. It can be applied to the quantification of biological and technological variation and to the identification of differentially expressed genes in two or more conditions.http://www.bioinformatics.org/project/?group_id=465

Answer 3

[Priyanka Surana]

Thank you everyone. They were very helpful comments and answers. I have a good starting point to go from. :) Priyanka