Go terms enrichment analysis

Question

[c.bessem]

I generated an annotated transcriptome of a non model organism using Trinity and Trinotate. I would like to perform some GO terms enrichment analysis. I did this a month ago on the website of wego (http://wego.genomics.org.cn/) but it's not working anymore. All the other alternatives I checked (AMIGO, goProfiles ...) use proper gene annotation as starting point but I only have the GO terms associated with my transcript. Is there a way to do the analysis starting with the file I have.

Here is an examples of my files:

transcript GO term

TR35016|c140_g012080 GO:0005576; GO:0005319

TR8218|c4_g015662 GO:0005576; GO:0005319

TR9751|c8_g015821 GO:0005576; GO:0005319

TR27091|c6_g010849 GO:0016021; GO:0005743; GO:0070469; GO:0004129; GO:0020037; GO:0005506; GO:0009060

TR24288|c6_g010207 GO:0005743; GO:0045275; GO:0009055; GO:0046872; GO:0008121

TR8216|c3_g015661 GO:0005576; GO:0005319

TR35068|c18_g012086 GO:0005576; GO:0005319

TR31387|c2_g011562 GO:0005576; GO:0005319

Thank you for your help.

Answer 1

[Guangchuang Yu]

clusterProfiler supports using your own annotation for both hypergeometric test and gene set enrichment analysis.

BTW: wego was not updated since 2009. The data is out-dated.

[c.bessem]

thanks for the suggestion. Do you have an easy to understand manual for your package. I tried to look at the github page but the instructions are hard to understand especially that there are no examples of the input formats.

I know that WEGO is outdated but its main advantage is that its really easy to use.

[Guangchuang Yu]

If you read the paper, http://biorxiv.org/content/early/2016/04/19/049288, you will believe that WEGO needs to die.

Figure B shows that in 2009, there are about 10k terms with gene annotations for Human, while in 2016, the number is more than 20k. This is a huge impact and your enrichment result could be totally different.

[Farbod]

Yes you are right

but I guess it is still popular because you can create beautiful comparison graphs by just 2 - 3 click !

[Guangchuang Yu]

AND clusterProfiler can do such thing in also just 2-3 command.

[Farbod]

So it would be a chance for us. Thank you

[c.bessem]

Dear Guangchuang, Is there examples of the input files clusterProfiler needs?

[WouterDeCoster]

Read the documentation: http://guangchuangyu.github.io/2016/01/go-analysis-using-clusterprofiler/

[Farbod]

Dear Guangchuang Yu,

I am agree with @c.bessem,

your software seems very interesting and powerful, but it seems that it is for experts (which I am not one of them ;-) )

[WouterDeCoster]

clusterProfiler looks quite well documented: https://guangchuangyu.github.io/clusterProfiler/documentation/

Answer 2

[Farbod]

Hi Dear c.bessem,

Good news! the WEGO site is Open again.

http://wego.genomics.org.cn/cgi-bin/wego/index.pl

Answer 3

[h.mon]

If you used Trinity and Trinotate, you could run goseq as per their documentation.

Answer 4

[c.bessem]

@h.mon thank you for the suggestion but I already used goseq and the results were not satisfying. I wanted something that could be a replacement to wego. Searching better I found Plogo (an R package that is similar to wego). I will try it and let you know. My main objective is to compare the same tissue but from different species. If you have other tools in mind, please let me know (taking into account the type of input files I have)

[WouterDeCoster]

Please use ADD COMMENT to answer to earlier posts, as such this thread remains logically structured and easy to follow. Thanks!

[Farbod]

Hi Bessem,

why this famous and useful site is closed? Do you have any idea?

[c.bessem]

@ WouterDeCoster, I'll do it next time @ Farbod, I have no idea. it was working in August (the last time I used it). Anyway the Plogo library in R works just fine but without a statistics to support the analysis.

Answer 5

[Farbod]

Hi,

Have you even tried CateGOrizer or GOLEM ?

[c.bessem]

I tried both of them but they don't have the option to compare two samples

Answer 6

[Prasad]

have you tried agrigo. You can use your own background dataset for enrichment