I'm trying to identify miRNA targets using online tools such as Targetscan, Pictar, Diana, miRDB. I am trying to find a final target list (intersect of all sub lists) in order to make GO.
However,
1) I'm not sure about levels of context score I should use in each database.
2) Should I exclude the gene which is in the first three but not in the fourth one?
Strange that nobody has responded to this question.
Similar question have turned up here on Researchgate and generally the mixing of databases is not recommended. Why? There was a paper a few years ago that did not see any benefit to do so (unfortunately, I forgot the quote but a similar question here was answered referring to this publication!).
I generally go with Targetscan. I am old-fashioned. Most of the predicted targets will most likely be false positives anyways and Targetscan has a pretty good track record compared to other algorithms.
So, my short answer is: don't do it. If you need a guide check out this website: https://sites.google.com/site/mirnatools2/.
This is a good start to find confirmed target sites by going to the tarbase link or similar sites listed on miRnatools. But in my opinion, even in tarbase a re a lot of false positive targets listed. In the end you may have to do an unbiased approach and look for targets yourself.
This may not have been the answer you were looking for, but this is my view. Perhaps some alternative answer will show up here in a while...
Thanks a lot for your opinion. I have tried to combine predicted gene sets from Targetscan and Diana-microT-CDS. In case of 8 different miRNAs the intersection was from 10 to 40%. I decided to use Diana and Diana-mirpath, its simple, it is possible to upload list of miRNAs in the text file. Great function is also the pathway enrichment using Kegg terminology.
Dear Alena,
Please just remember that many of miRNA:transcript interactions may occur in CDS, so I would reccomend a tool/database that is accounting for that fact, I have used microT4-CDS and PACCMIT-CDS.
I am in favor of using at least few tools together to find more confident calls, but fixating on the fact that your putative regulation has to be found by all of your tools is probably wrong. My approach is to use 4 similar tools and selecting putative regulations that were confirmed by at least 2 out of 4 tools. However I don't know if that is a best approach,
Best of all,
Bartosz Wojtas
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Biomolecules
- Nucleic Acids
- RNA
- Small RNA
- microRNA