How to cluster GO pathways well?

Question

[kim]

I have an issue about clustering of gene ontology pathways.

In a nutshell, I obtained some differential expressed genes and did gene ontology enrichment analysis using GOstats. And I may do further analysis for functional clustering and visualization including Cytoscape.

Simply before taking another step, I'd like to drawing a simple plot like this (Fig1A. http://www.ncbi.nlm.nih.gov/pubmed/20926834). In Fig1A, y axis represents negatively transformed p-value of enriched gene sets and specially, gene sets on x axis are displayed and clustered by effect size.

I've also emailed to the author about calculation of effect size of specific gene sets and clustering them. But I've gotten no reply from him. What should I do to cluster gene sets simply not tricky (Just like this figure!). Do you have any idea or recommendation?

Answer 1

[russhh]

Hi,

From your question I can't be sure what you've attempted so far. Does the following help?

df <- data.frame(
  effect = rnorm(100),
  pval = rbeta(100, 0.1, 10))
hc <- hclust(dist(df$effect))
ord <- hc$order
plot(-log10(df$pval[ord]), xlab = '', ylab = 'pvalue (-log10)')
plot(as.dendrogram(hc), main = '', xlab='')