microarray differential expression analyses

Question

[bioinformatics]

Hi,

I have performed microarray differential expression analyses and generated the following table of top 10 DEGs.

I just had a question if it were possible to have the same gene symbol e.g. EEF1A1 repeated 3 times in the table. Are they different isoforms of the same gene?

Please find below the table.

Gene symbol Probe ID    Log FC  AveExpr t   P value adj.P.Val   B
HUWE1   207783_x_at 13.65029965 13.6707791  893.2391539 5.76E-183   5.76E-183   387.6583082 TPT1    212869_x_at 14.1482378  14.1966412  800.8523828 1.32E-178   1.32E-178   381.4258321
N/A AFFX-hum_alu_at 14.56974087 14.54781421 798.9384531 1.65E-178   1.65E-178   381.2834173
EEF1A1  213477_x_at 13.58424538 13.62768221 723.7812901 1.46E-174   1.46E-174   375.1934051
TPT1    212284_x_at 13.70845455 13.76689133 695.5901055 5.64E-173   5.64E-173   372.6324936
RPL23A  208834_x_at 13.90989102 13.96648201 687.5728051 1.64E-172   1.64E-172   371.8735615
RPL41   201492_s_at 14.20547628 14.24722228 686.0505397 2.01E-172   2.01E-172   371.7278708
EEF1A1  206559_x_at 14.03777874 14.09315376 664.84566   3.61E-171   3.61E-171   369.6439859
EEF1A1  204892_x_at 14.17778189 14.23010715 664.5006272 3.78E-171   3.78E-171   369.6092185
RPS3A   212391_x_at 13.63254758 13.65580573 622.69442   1.49E-168   1.49E-168   365.179012

Thankyou

Answer 1

[ATpoint]

It's different probes per gene, can be isoforms or splice variants indeed.

https://support.bioconductor.org/p/92128/

These logFCs in your table are suspiciously high, so are the significances. If you want you can show some code, I have the gut feeling something is very wrong here.

[bioinformatics]

Thanks for your feedback. Please find my r script below:

> library("arrayQualityMetrics")
> library(GEOquery)
> library(oligo)
> library(Biobase)
> library(affy)
> library("splitstackshape")
> library("tidyr")
> library(dplyr)
> celFiles <- list.celfiles()
> affyRaw <- read.celfiles(celFiles) Platform design info loaded. Reading in : GSM766537.CEL Reading in : GSM766539.CEL Reading in : GSM766624.CEL Reading in : GSM766640.CEL
> eset <-oligo::rma(affyRaw) Background correcting Normalizing Calculating Expression
> library(limma)
> pData(eset)
              index GSM766537.CEL     1 GSM766539.CEL     2 GSM766624.CEL     3 GSM766640.CEL     4
> Groups <- c("DDLPS", "DDLPS", "WDLPS", "WDLPS")
> design <- model.matrix(~factor(Groups))
> colnames(design) <- c("DDLPS", "DDLPSvsWDLPS")
> fit <- lmFit(eset, design)
> fit <- eBayes(fit)
> options (digits =2)
> res <- topTable (fit, number = Inf, adjust.method = "none", coef = 1)
> write.table(res, "diff_exp.txt", sep= "\t")
> require(hgu133a.db)
> probes <- rownames(eset)
> annotLookup <- select(hgu133a.db, keys = probes,
+   columns = c('PROBEID', 'ENSEMBL', 'SYMBOL')) 'select()' returned 1:many mapping between keys and columns
>

Please let me know if I have made an error. I did use all the samples (52 WDLPS and 40 DDLPS) not just the 4 samples as shown in this script.

[ATpoint]

Yes, it seems wrong indeed, you need coef=2 to do differential expression. With coef=1 you are testing whether the intercept is zero and that has no meaning and gives meaningless stats, see https://support.bioconductor.org/p/85730/#85740 and be sure to not go blind. Read the limma user guide!

[bioinformatics]

Yes, ok thanks for this correction. I will re-do the analysis and change coef=1 to coef=2.

Do you know if the the rest of the script is correct?

[ATpoint]

I will not guarantee for it -- see the limma manual, it covers everything you need.