Gene ontology comparsion between up- and down-regulated genes

Question

[seta]

Hi all friends,

I did differential expression analysis of an RNA-seq project. Then, using agrigo (enter link description here), I got enriched GO terms for up- and down-regulated genes, separately. Now, I was wondering how to statistically compare these GO terms between up and down regulated genes and show the results on the graph?

Thank you

[EagleEye]

Option 1:

Try to present in this way, instead of microarray and RNAseq you can use up and downregulated category.

Figure source from Supplementary Figure 4. http://www.nature.com/ncomms/2015/150724/ncomms8743/extref/ncomms8743-s1.pdf

OR

Option 2:

http://genescf.kandurilab.org/ftp/GeneSCF-Documentation_v1.1.pdf

Page 12 (ggplot2 bubble plot)

OR

Option 3:

If you do not want to show number or percentage of genes appeared in each term, you can use simple double bar graph (up and down) with log10(Pvalue).

[seta]

Thank you very much for your prompt response. The option 1 is great, but this is my first experience. Could you please help me out how I can draw a graph similar to this graph?

[EagleEye]

It is bit tricky to explain, but I will try my best.

1 Make your GO for up and downregulated as simple table

GO_name <tab> log10(p)_UP <tab>  log10(p)_DOWN <tab> UP_genes <tab> DOWN_genes

GO1 <tab> xx <tab> xx <tab> xx <tab> xx
GO2 <tab> xx <tab> NA* <tab> xx <tab> NA*

*NA represents not enriched in one of the category

2 Make heatmap for the log10 pvalue columns in the table using R heatmap.2

3 Make bar graph for the number of genes columns in the table using barplot

4 Use Inkscape (FREE) / Adobe Illustrator to place bar graph over heatmap.

[seta]

Many thanks for your explanation. I'll try it. Just one thing about GeneSCF that you also suggested, as I mentioned in my post, I have enriched GO terms for up and down-regulated genes. Would you please let me know if I just show them, without doing any statistical analysis (because I did it before) by this tool?

[EagleEye]

It is better to show or present it with proper statistics. This will gain more confidence in your data. And for your information GeneSCF provides results with extensive statistical analysis.

Answer 1

[asanchez]

Hello I have just found this post You might consider using the goProfiles Bioconductor package (http://www.bioconductor.org/packages/release/bioc/html/goProfiles.html) whic is precisely intended to compare two gene lists based on their annotations at a geiven GO level. I am the maintainer of the package so feel free to ask any question if you want