16
$\begingroup$

As enrichment analysis a usual step is to infer the pathways enriched in a list of genes. However I can't find a discussion about which database is better. Two of the most popular (in my particular environment) are Reactome and KEGG (Maybe because there are tools using them in Bioconductor). KEGG requires a subscription for ftp access, and for my research I would need to download huge amounts of KGML files I am now leaning towards Reactome

Which is the one with more genes associated to pathways ? Which is more completely annotated ? Is there any paper comparing them ?

Improve this question
$\endgroup$
0

3 Answers 3

Reset to default
4
$\begingroup$

One big downside of KEGG is the licensing issue. One big advantage of Reactome are various crosslinks to other databases and data.

ad 1, This depends on which pathway, they are both primary databases. Sometimes other databases that for instance combine data of primary databases have better annotation of pathways (there is an example in the review paper bellow)

ad 3, There is very extensive relatively new (2015) review on this topic focused on human pathways: Comparison of human cell signaling pathway databases—evolution, drawbacks and challenges. However I could not find there which one is more complete ...

Improve this answer
$\endgroup$
4
$\begingroup$

For you the main point would be whether an enrichment analysis is going to give you an informative answer. That's what will make a particular database better. And there's all sorts of subjective decisions made in their construction as to what what a pathway is, what to include, where to draw boundaries, etc. so different databases will give different answers and it will not be clear which is more correct.

So browse over the references / pointers people have given, select a service and use it and no others. Don't jump around databases until you get an answer you like, that's just fishing.

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ +1 for the last sentence, is so easy to say I will check with X once one have a "good" results and if both agree then go on. $\endgroup$
    – llrs
    Commented May 26, 2017 at 10:32
1
$\begingroup$

One big advantage of Reactome, in my opinion, is its visualization using the web interface.

Many pathways (in Reactome and KEGG) consist of genes / proteins that are up- and down-regulated through the respective pathway. If you do a simple overrepresentation analysis this is not taken into consideration. Therefore, you might end up seeing a pathway as "overexpressed" although only the down-regulated genes were observed more frequently.

In Reactome, you can zoom in on the different pathways very conveniently and then pick up these inconsistencies. I haven't really found a public database and tool that can take these different regulations into consideration. Therefore, you'll probably always will need some manual investigation of your data. In my opinion, this is easier with Reactome than with KEGG.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.