Finding overlapping genes between different gene sets

Question

[kim]

I have an issue calculating overlap percentage.

From functional annotated analysis, I got 49 different gene sets and have tried to calculate how many genes are overlapped between two different data sets. I'd like to drawing a figure like this (Fig1B). Are there any ideas or recommendations?

http://www.ncbi.nlm.nih.gov/pubmed/20926834

Actually, I am still new to R but prefer to R.

Answer 1

[Jason]

It's difficult to tell what problem you are having in reference to calculating overlap. If you could edit your post to add more info you may get more help. For instance, show us some of the data and what you want the output to look like.

When it comes to drawing heatmaps there are previous posts on that on this site related to that. I'd also check out the following packages/tutorials.

• This post from stack overflow (actually produces the half heatmap you want)

The following links make regular square heat maps but you could easily make the figure like the one you linked by importing the figure into photoshop, powerpoint, etc and editing out half of the heatmap.

• Here is the heat map package in R
• Heatmap tutorial
• Heatmap.2 tutorial
• Heatmap with ggplot2

[kim]

Thank you so much for your help :)

Actually I obtained 49 enriched gene sets using GOstats R. Plus, I have annotated gene lists of each 49 data sets. Here, I'd like to figuring out 1) how each category share same genes together and/or 2) which genes are annotated on different categories among total 49 sets. GO datasets are displayed with GOBP ID and genes are expressed with ENTREZ ID.

   GO.0071779   GO.0006091  GO.0031570  GO.0051179
1       5682         32       4436         32
2       5684        226       5682         92
3       5693        498       5684        116
4       5700        506       5693        161
5       5705       2026       5700        226
6       5706       2271       5705        372

Following your recommendations, I have plotted different heatmaps using function heatmap and image. But.. these gave me a little different format what I expected before (figure from published paper I'd attached earlier). But, I'm trying to make one..

[Mo]

It is very interesting, if you find the way to do it, please also share it here.