Question

Meta-Analysis after running limma on a normalized dataset

0

Entering edit mode

Vani ▴ 20

@vani-8145

Last seen 8.9 years ago

United States

Hi,

I am trying to run a meta-analysis on multiple datasets. After the normalization was completed I applied, limma's lmfit,ebayes, and toptable on the datasets. I would like to use the data produced by toptable to do a meta-analysis on the datasets.

This is the header of my top tables:

[1] "ENTREZID" "SYMBOL" "GENENAME" "logFC" "CI.L" "CI.R"

[7] "AveExpr" "t" "P.Value" "adj.P.Val" "B"

After performing a meta-analysis I would also like to do a forest plot on the result.

What would be the best way to do this?

Thanks

meta-analysis limma toptable • 3.0k views

ADD COMMENT • link updated 9.4 years ago by Aaron Lun ★ 28k • written 9.4 years ago by Vani ▴ 20

0

Entering edit mode

I am not sure if it makes sense to do traditional meta-analysis on gene expression data. What I have found is that using packages such as RankProd works pretty well. You would take the directional t-statistics from limma, make a table out of them and then do one-sample RankProd analysis (or Ranksum if you like). Maybe you can do the forest plot afterwards on the few remaining genes. RankProd can deal with missing values (or you could do imputation also).

ADD REPLY • link 7.9 years ago Pekka Kohonen ▴ 190

score 2 · Answer 1 · 2015-07-21

2

Entering edit mode

Aaron Lun ★ 28k

@alun

Last seen 19 hours ago

The city by the bay

I don't think that there are any functions in limma for constructing forest plots. You'll have to use some other packages; a quick Google search brings up forestplot and rmeta. With the former, it seems like you can make a forest plot for each gene, by supplying the log-FC with the upper and lower bound for the confidence interval. Make sure you've matched up the genes properly between data sets, though.

ADD COMMENT • link 9.4 years ago Aaron Lun ★ 28k

0

Entering edit mode

Hi,

Before I create a forest plot I want to run a fixed and random effect meta-analysis process across all the 4 datasets. I couldn't find a way to do this on rmeta? Do you have any advice in this regard?

ADD REPLY • link 9.4 years ago Vani ▴ 20

0

Entering edit mode

Look into the meta and the metafor packages for this. The metafor package has an extended website which might be more informative for a quick overview of what type of functionality it might provide you

ADD REPLY • link 9.4 years ago Steve Lianoglou ★ 13k