I have looked on the web for gene set enrichment analysis tools with which to evaluate the results of my feature selection work and I found a world of alternatives.
For me it is not easy to choose one or another, so which is your favorite platform and why?
The objective of my work is to compare different gene lists from microarray experiments related to leukemia or other cancer types. I would like to know which list is best, considering the disease which is being studied.
Hi Mattia,
After searching and using different tools for gene enrichment analysis, I settled on the very user-friendly Enrichr, there : http://amp.pharm.mssm.edu/Enrichr/
and its accompanying framework, L2N : http://amp.pharm.mssm.edu/lachmann/upload/login.php.
In my hands I find those more practical and easy to use than classical DAVID website/tool, and more sensitive, too.
On the plus side :
* User-friendly interface
- you can choose to load your gene list once directly on Enrichr (by simple copy and paste of the official gene symbols), and have all the results at once, or you can load the same list into L2N first (also by copy & paste, or by uploading a larger files), and keep it for later analysis.
- the lists loaded via L2N are kept (indefinitely, from what I can tell) to be used later, protected (by login) and can be shared between users.
- there is a very handy direct link for each list allowing its direct analysis into Enrichr
- the web-page are clear, simple, and easy to deal with.
- you can correct a gene symbol if you have pasted the non-offical one (i.e a synonym)
- lists can be large (at least up to 15 000 genes, in my hands)
* Complete and simple analysis
- Enrichr allows you to search for enrichments in all categories at once : Transcription (with 6 items, including ChIP data and miRNA targets), Pathways (9 databases, with KEGG, of course ;-D), Ontology (4 items, GO-BP being the most useful for me), Disease/Drugs (8 items, including OMIM), Cell Types (4 items), Miscellanious (4 items, with a useful "chromosomal location").
- all the results are presented in tabs, and open in the same window
- for each enrichment, you can have different presentation : a visual bar graph (that you can sort by three different criteria, I always use p-values, and with user-chosen colors), a table (that you can sort also, and download to keep the results), a network for related enriched terms, and a grid, looking like a microarray matrix.
IMPORTANT : all p-values presented directly are already corrected with bonferroni correction, meaning that what you see enriched is REALLY significant. However, when you download the tabular results, you have both uncorrected p-values and bonferroni-corrected p-values in the table.
- from the L2N interface, you can work with your lists, computing their overlap, intersection, union or exclusion.
- the sharing of genes between lists is depicted with a very visual matrix, and comes with computed p-values for the significance of the sharing.
- Also, you have the possibility to test the enrichment of several lists at once for the same item, i.e. KEGG pathways for 20 lists together. The matrix will show you how your 20 lists are connected and what are the respective enrichments for pathways.
On the minus side :
well, not many...
- in L2N, the network module is not very helpful for me at the moment, and could be really improved, in particular if one could choose the items to be pictured and deal with fonts, colors, size, etc...
- in L2N again, I miss a way to organize my different lists. For the moment, they are just listed in uploaded order (the most recent first).
- the manual correction of unrecognised Gene symbol is fastidious when a long list is copied, and the automatic correction tool that is available is not 100% correct (I've dealt with this issue by reannotating my transcriptomes).
Apart from these minor problems, these are great tools, and I use them every day !!
(and no, they are not mine... ;-D)
Hope this will help, all the best.
PS : I'm working with human genes, I have not tested these tools with other species.
Hi Mattia,
After searching and using different tools for gene enrichment analysis, I settled on the very user-friendly Enrichr, there : http://amp.pharm.mssm.edu/Enrichr/
and its accompanying framework, L2N : http://amp.pharm.mssm.edu/lachmann/upload/login.php.
In my hands I find those more practical and easy to use than classical DAVID website/tool, and more sensitive, too.
On the plus side :
* User-friendly interface
- you can choose to load your gene list once directly on Enrichr (by simple copy and paste of the official gene symbols), and have all the results at once, or you can load the same list into L2N first (also by copy & paste, or by uploading a larger files), and keep it for later analysis.
- the lists loaded via L2N are kept (indefinitely, from what I can tell) to be used later, protected (by login) and can be shared between users.
- there is a very handy direct link for each list allowing its direct analysis into Enrichr
- the web-page are clear, simple, and easy to deal with.
- you can correct a gene symbol if you have pasted the non-offical one (i.e a synonym)
- lists can be large (at least up to 15 000 genes, in my hands)
* Complete and simple analysis
- Enrichr allows you to search for enrichments in all categories at once : Transcription (with 6 items, including ChIP data and miRNA targets), Pathways (9 databases, with KEGG, of course ;-D), Ontology (4 items, GO-BP being the most useful for me), Disease/Drugs (8 items, including OMIM), Cell Types (4 items), Miscellanious (4 items, with a useful "chromosomal location").
- all the results are presented in tabs, and open in the same window
- for each enrichment, you can have different presentation : a visual bar graph (that you can sort by three different criteria, I always use p-values, and with user-chosen colors), a table (that you can sort also, and download to keep the results), a network for related enriched terms, and a grid, looking like a microarray matrix.
IMPORTANT : all p-values presented directly are already corrected with bonferroni correction, meaning that what you see enriched is REALLY significant. However, when you download the tabular results, you have both uncorrected p-values and bonferroni-corrected p-values in the table.
- from the L2N interface, you can work with your lists, computing their overlap, intersection, union or exclusion.
- the sharing of genes between lists is depicted with a very visual matrix, and comes with computed p-values for the significance of the sharing.
- Also, you have the possibility to test the enrichment of several lists at once for the same item, i.e. KEGG pathways for 20 lists together. The matrix will show you how your 20 lists are connected and what are the respective enrichments for pathways.
On the minus side :
well, not many...
- in L2N, the network module is not very helpful for me at the moment, and could be really improved, in particular if one could choose the items to be pictured and deal with fonts, colors, size, etc...
- in L2N again, I miss a way to organize my different lists. For the moment, they are just listed in uploaded order (the most recent first).
- the manual correction of unrecognised Gene symbol is fastidious when a long list is copied, and the automatic correction tool that is available is not 100% correct (I've dealt with this issue by reannotating my transcriptomes).
Apart from these minor problems, these are great tools, and I use them every day !!
(and no, they are not mine... ;-D)
Hope this will help, all the best.
PS : I'm working with human genes, I have not tested these tools with other species.
Thanks to Sandrine for telling us about Enrichr. I am a long-time DAVID user, but having tried Enrichr, I really like many its features, particularly the easy way of selecting different data sources and its very clean interface. One short fall I saw in comparison with DAVID is its sole support for gene ID and requirement for a clean gene ID list.
There is not a perfect tool for this. I am also looking for the answer of this question.
Take a look on this paper might be it is easy from you to identify the best tool for enrichment. http://www.ploscompbiol.org/article/info%3Adoi%2F10.1371%2Fjournal.pcbi.1002375
But i couldn't able to make a decision yet about the best enrichment tool.
I sometimes use the gene set analysis functionality within Babelomics: http://babelomics.bioinfo.cipf.es/functional.html. Why: mostly because it is easy and quick to use and so far I did not have any problems with it. Anyhow, I don't use it too often, as typically I apply simple statistical overrepresentation tests in R.
Interesting approach which we are using is GO-Bayes (http://bioinformatics.oxfordjournals.org/content/26/7/905).
However, as Katrin has said for quick check what is going on Babelomics is good enough.
Hi all, I also endorse the use of Enrichr for many reasons:
1. Online HTML5 super easy and reeeally fast.
2. Plenty of Gene Category Databases, even newest, Roadmap Epigenomics, ENCODE
3. Z-score Permutation Background Correction on FisherET p-value specially for large genesets
4. KEGG 2015!!! and keeping all databases updated (DAVID???)
5. API and programmatic access which gives you the opportunity to attach enrichr enrichments to your favourite pipeline of analysis and evaluate multiple enrichments without multiple clickings rather easily.
Hope the guys at Mayan's Lab keep on being funded in the future to sustain such a nice tool free.
PS. Also Toppgene suite is worth trying.
I recently checked out Enrichr, and it is easily the easiest to use, prettiest, and apparently the most updated, However, it lacks the ability to upload a background gene set or BED file, so if the background of your experiment is not genome-wide, then Enrichr may not be the best tool. I saw in their most recent paper that Enrichr people are currently working on a function to allow uploading background data, but until they do, I will be staying with GREAT because my experiment is not fully genome-wide.
Agreed with some comments above, many enrichments tools don`t give proper background set-up, I little prefer David and Goseq (R package) now since they allow me to upload "custom" background for my enrichement test. But the main, perhaps the best feature of EnrichR compare to other webtools is the reproducibility of analyses since you can save and share your analyses.
Hi. Eventhough its a pretty old question, just thought of adding some information to this.
Recently, I had tried using several tools for Enrichment analysis. I found, Consensus Database to be very user friendly and reliable. It also allows us to upload our background gene list and uses 11 different databases (which can be selected based on our interest)
Apart from that it has features for:
- Metabolite analysis
- Shortest Interaction paths
- Molecule or pathway interactions
Link:
http://cpdb.molgen.mpg.de/
PANTHER is a very useful web tool.
http://www.pantherdb.org/
An updated list of other suitable tools can be found here:
http://www.molbiotools.com/usefullinks.html#fungen
As said several times already, the perfect tool allows you to correct for background. At the moment I know both DAVID and STRING can do that (and possible others too, I just have no experience with them). Enrichr as far as I can see not yet. Genome-wide background when having limited detection (biological and/or technical reasons) would result in very biased results. Have a read here on the topic: Article Multiple sources of bias confound functional enrichment anal...
Enrichr, no doubt. I have used Panther, David and gProfiler and Enrichr is the best compromise between simplicity and comprehensiveness... Its API is just awesome, you can write scripts to automate analyses.
HI there
Does anyone have an opinion on metascape? I'm intrigued as to why this is not mentioned as so easy to use?
Thanks
Amy
- Badges
- Science topic
- Similar topics
- Biological Science
- Genetics
- Genetic Analysis