functional enichment with custom gene annotation

Question

[zhang.jianhai]

Hi,

I have a list of gene-annotation pairs and want to do enrichment analysis on specific sets of genes. There are many functional enrichment tools, but few allow me to input my own gene-annotation (each gene belongs to a network) info for the enrichment. What are the tools that can be used in a plug-and-play manner? Preferably, it provides multiple ways of visualisation on the enrichment output.

Thanks.

[Lluís R.]

You are commenting that you have a list of genes that are networks. How do you define those networks ? And what type of enrichment are you looking for?

[zhang.jianhai]

They are defined according to their expression profiles measured by RNA-seq. I want to enrich memberships in some gold standard networks so as to understand how my own networks reflect reality.

[Lluís R.]

You keep using the word network but you don't specify in which format do you have this network. How are the edges and nodes of this network stored? Or these custom network is just a list of co-expressed or a selection of genes?

Answer 1

[Jean-Karim Heriche]

You can use the R Bioconductor package topGO with custom annotations which I think you can use with the ViSEAGO package for visualization.

[zhang.jianhai]

On the tutorial of topGO (https://bioconductor.org/packages/release/bioc/vignettes/topGO/inst/doc/topGO.pdf). It seems this package is only applicable to GO enrichment. My annotation is about which gene belongs to which network (I did not state clearly my question at the beginning). So topGO is not applicable in this case, right?

[Jean-Karim Heriche]

Think of each network as a pathway/biological process. You simply annotate each gene with the network(s) it belongs to. If there's no relation between the networks then this doesn't have the complexity of dealing with the GO hierarchy of terms. So if your question is about enrichment in membership in some networks, you just need to build a contingency table for each network and do a chi-squared test and correct for multiple testing. If that's not it, then you need to explain more about what you're trying to do.

[zhang.jianhai]

Yes, my question is about enrichment in membership in some gold standard networks. I constructed many networks, and each needs such enrichment on gold standard networks, so it should be a programmatic process. I like the powerful visualisations in some GO enrichment tools such as clusterProfiler, so I wonder if some of them can input gene-network annotations for enrichment. If not, I have to build the enrichment and visulisation pipeline by myself.

Answer 2

[grant.hovhannisyan]

clusterProfiler is also a nice Bioconductor package https://bioconductor.org/packages/release/bioc/vignettes/clusterProfiler/inst/doc/clusterProfiler.html

[zhang.jianhai]

There is instruction (https://yulab-smu.github.io/clusterProfiler-book/chapter5.html) of using clusterProfiler on custom annotations, but it only allows custom GO annotations. My custom annotation is predefined networks, e.g. each gene belongs to a network.

Answer 3

[padwalmk]

Hi, If you have got your answer than ok, otherwise you can try Cluego plugin of the Cytoscape. It requires a license but once you register you will get a license. It's pretty easy and helpful for gene-ontology analysis. It can take direct input from your network, and in addition, it has a feature for the gene-ontology term fusion. It allows not only the gene-enrichment but pathway from KEGG, Reactome, and immune system-related. Different layout for visualization, check out particularly cerebral layout which gives you a real feel for cell organization with terms and genes labeled in a hierarchical manner.