In clusterProfiler: statistical analysis and visualization of functional profiles for genes and gene clusters

How to use clusterProfiler to do GO enrichment analysis with unsupported organisms

Guangchuang Yu

Jinan University

This vignette is an extension for what already exists in the clusterProfiler.pdf vignette. The clusterProfiler provide enrichGO function to do hypergeometric testing with “human”, “mouse” , "zebrafish" and “yeast” organism supported. It is very easy to support other organism provided that the bioconductor annotation package exists.

Most of the software packages for GO enrichment analysis in the Bioconductor project were designed for model organism, and they all rely on the bioconductor annotation packages.

If the organism without annotation package available, it is not easy to employ the existed package to perform such an analysis.

I have extended clusterProfiler to support the unsupported organism.

Here, I will illustrate how to do GO analysis for Streptococcus pyogenes M1 MGAS5005, as an example.

For doing GO analysis, you should have gene and GO mapping data.

I suppose you have nothing in hand, and explain how you get these things in hand.

The whole genome annotation can be downloaded from NCBI. In this example, the M5005 bacteria whole genome annotation file can be downloaded from: ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Streptococcus_pyogenes_MGAS5005_uid58337/

The clusterProfiler package has functions for parsing GFF file. When you have downloaded the gff file, using the following command:

suppressPackageStartupMessages(require(clusterProfiler))
Gff2GeneTable("NC_007297.gff")

This function will parse the gff file, and extract information to form a data.frame, and save it as “geneTable.rda” in the current working directory.

load("geneTable.rda")
head(geneTable)

This geneTable is useful for ID mapping, and will be used for mapping GeneID to GeneName if parameter readable is set to TRUE when calling enrichGO.

eg <- geneTable$GeneID

Now you have all GeneID stored in eg, I recommend you using biomaRt package to query GO annotation, and I will demonstrate how to do it.

require(biomaRt)
bacteria=useMart("bacteria_mart_16")
bac = useDataset("str_22007_gene",mart=bacteria)
gomap <- getBM(attributes = c("entrezgene",
               "go_accession"),
                filters = "entrezgene",
                values = eg, mart = bac)

dim(gomap)
head(gomap)

You should use other dataset for other bacteria. If the organism is not bacteria, you should use other mart, for instance fungi_mart_13 for fungi.

The gomap only contain GO directly annotation, but undirectly annotation was needed for GO enrichment analysis.

So, clusterProfiler provided another function called buildGOmap, for building gomap files needed for analysis.

buildGOmap(gomap)

After running this command, buildGOmap function generate GO2EG.rda, EG2GO.rda, GO2ALLEG.rda and EG2ALLGO.rda in the working directory.

Providing these files in the working directory. The enrichGO function can perform hypergeometric test for this organism.

Suppose the following genes are of interested.

gene <- c("3572890","3572609","3572407","3572408","3572333",
          "3572206","3572193","3571922","3571782","3571786",
          "3571624","3571626","3571412","3571413","3571382",
          "3571286","3571289","3571124","3571106","3571029")
mf <- enrichGO(gene, ont="MF", organism="M5005", pvalueCutoff=0.05, readable=TRUE)
summary(mf)

You can used other tools provided in clusterProfiler, such as plot to visualize the result, and compareCluster to compare different gene clusters.

clusterProfiler documentation built on Feb. 11, 2021, 2:02 a.m.