Gene enrichment tools options
1
0
Entering edit mode
3 months ago
Sahara • 0

Hi hello, A newbie here in the bioinformatics research. I'm currently doing a research on microRNA as PTC (papillary thyroid carcinoma)'s diagnostic biomarker and I found some difficulties in the gene enrichment step for my microRNA candidates target genes. I have already tried to use ShinyGO,DAVID, and Human Mine but all of them provided unsatisfying results or even no results at all for some of the gene list. Are there any other gene enrichment webtools that I can try? or Is the probem lie within the gene list? And IF it is, what should I do to fix it?

Thank you for anyone who read or even willing to answer this, I'm sorry in advance if my questions or sentences doesn't make any sense. Thanks once again for helping a student in despair, have a nice day ^^

DAVID ShinyGo gene-enrichment • 730 views
ADD COMMENT
0
Entering edit mode

Why is it unsatisfying ? How many genes are in your list ? What kind of genes (protein coding only or also non-coding) are in your list ? What kind of gene ID/gene names do you have in your list (ENSEMBL, ENTREZ, etc...) and do you know whether those genes ID are recognized by the tools you used ?

Answering those questions will help you (and us) determine whether something went wrong or if having no result is normal.

ADD REPLY
0
Entering edit mode

Hello, thank you for replying my questions. First of all i have three lists of genes consisting of 4 (A), 18 (B), and 16 (C) genes. I have no problem whatsoever with the A gene list, the one that only contains 4 genes, the problem only arise with the B and C gene list.

  1. It's unsatisfying because it doesn't show any result at all at every webtools that i used for the B and C gene list , even though i think it had a pretty good numbers of genes(I'm sorry if my choose of words was offensive)
  • For example, in Shiny GO, when i submitted the B and C gene sets, the results are "Error: an error has occured".
  1. I think all of them are coding genes? because based on DAVID's analysis, each of the genes has its own david gene name.

  2. I use ENSEMBL gene ID

  3. All of the genes that i use are recognized by the tools that I used

Thank you once again for replying

ADD REPLY
2
Entering edit mode

It's still unclear to me whether all of the tools emit an error, or just no significant results. If the latter, that may just be a reflection that you don't have any enrichment, which is normal for such small lists. If anything, I'm more surprised you got a significant result in an enrichment analysis with just 4 genes.

ADD REPLY
2
Entering edit mode

I agree that the probability to have no enrichment for less than 20 genes is relatively high. It obviously depends of how strong your signal is (if all the 20 genes are from the very same pathway, then you will get enrichment), but in most cases, you might not have enough power to detect a statistically significant enrichment. In other word, medium-sized enrichment from such small list could easily occur by chance -> they are not considered statistically significant -> they are not reported.

If your goal is only exploratory, then I would suggest to take a descriptive approach called a functional profiling: regardless of the enrichment, what are those genes in your list, and to which biological function do they relate ? To do that, you could use gprofiler for instance.

ADD REPLY
1
Entering edit mode

A robust way to check small to medium sized results (and one I've had to do in peer review for enrichment analyses) is a circular genomic permutation analysis (first described here). You can then add your result into a one-tailed p-value distribution to check significance. It's impressive how often "significant" enrichment results come up when randomly sampling the transcriptome.

I second gprofiler. It's a great tool for exploratory analysis, especially in non-model systems (though that doesn't apply here).

ADD REPLY
0
Entering edit mode

Does circular genomic permutation analysis requires any spesific webtools or applications? Thank you for the suggestion and reference

ADD REPLY
1
Entering edit mode

No, but the authors have made an R package. I wrote a script to do it myself once I understood how it worked. But if you want a quick and dirty analysis for exploration, you can just do a normal permutation analysis randomly sampling n genes and performing an enrichment analysis 1000 times.

To my knowledge there isn't a referenced analysis showing the effect of sample size. But you can check the literature and see what sample size people in your field normally report.

ADD REPLY
0
Entering edit mode

I see, maybe the problem is really because of the amount of genes. Are there any journals or articles that I can refer to for this? And thank your for the suggestion, I'll take it into consideration

ADD REPLY
1
Entering edit mode
3 months ago

GO's official recommendation is the PANTHER tool: pantherdb.org. There is a help guide as well as numerous PANTHER publications. As others have mentioned, very small gene sets may just not have significant results.

ADD COMMENT

Login before adding your answer.

Traffic: 1638 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6