Question

Collapse FDR value of probes detecting the same gene

0

Entering edit mode

tofukaj • 0

@tofukaj-9928

Last seen 4.4 years ago

Dear all,

I'm very new in Microarray analysis, so I hope my question wouldn't be too irritating to you all.

I have pre-processed and filtered my microarray datasets and perform differential expression analysis. I would like to note that I do not consolidate my probes.

I have two probes detecting my interested gene, so I have 2 FDR values. I want to collapse them to acquire only 1 FDR value for my gene (and still keep the FDR value of each probe).

In brief, I want to combine two probe FDR values acquired from the same test in the same datasets (I hope I do not make any mistake here)

Would you please suggest me if this is possible and how to do it?

Thank you very much in advance,

Kaj

R version 3.2.4 (2016-03-10)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 14.04.4 LTS

locale:
[1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
[3] LC_TIME=th_TH.UTF-8        LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=th_TH.UTF-8    LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=th_TH.UTF-8       LC_NAME=C
[9] LC_ADDRESS=C               LC_TELEPHONE=C
[11] LC_MEASUREMENT=th_TH.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] stats graphics grDevices utils datasets methods base

test fisher exact EmpiricalBrownsMethod affy • 1.4k views

ADD COMMENT • link updated 8.7 years ago by Gordon Smyth 51k • written 8.7 years ago by tofukaj • 0

score 2 · Accepted Answer · 2016-03-16

2

Entering edit mode

Gordon Smyth 51k

@gordon-smyth

Last seen 16 minutes ago

WEHI, Melbourne, Australia

You can't collapse FDR values. If you want to consolidate duplicate probes for the same gene, then this needs to be done earlier in the analysis. At very least, you need the p-values.

How exactly have you obtained the FDR values? The sessionInfo() you give suggests that you are not using any Bioconductor packages at all.

Also, is there a reason why you need to combine probes for the same gene? Most people do not bother combining probes if all they want is a list of differentially expressed genes.

ADD COMMENT • link 8.7 years ago Gordon Smyth 51k

0

Entering edit mode

Thank you very much for your response. First of all, I would like to apologize about sessionInfo(). I perform the differential test using 'lpe' package, and I corrected the FDR value by resampling FDR method.

The reason I need to combine the probe to the same gene is as indicated in another question of mine (that no one seems to be interested in it at all).

http://stats.stackexchange.com/questions/201520/combining-fdr-value-acquired-from-different-microarray-probes

As you can notice in my example, I have two probe giving different FDR for each paired test. I simply want to have another FDR value as overall representative value of the gene.

At this point, I'm going through probe consolidation as you suggest (so have another whole bunch analysis).

However, I have noticed that you mention about p-value as the least I required. As far as I can read (and poorly understand), I can only think of Fisher Exact test and Empirical Brown method. But none of them seems to fit my condition at all. Would you please suggest me or correct me (if I mistake something) about the test to combine p-value in my condition ?

Thank you very much in advance.

Kaj

ADD REPLY • link 8.7 years ago tofukaj • 0