[BioC] question about CORNA

[michael watson (IAH-C)]

Hi Maura

> It looks like CORNA is not another pair (miRNA - target gene) prediction program like miRANDA and so on ...  ?
You are correct, CORNA does not predict miR:gene relationships

> If I am guessing right (am I ... ?) then what is CORNA basing its miRNA -target_gene affinity score on ?
It can download targets data directly from miRBase targets, microRNA.org.  Alternatively, you can load your own data into CORNA as long as it is in tabular form

> Is it a synthesis of the predictions of the many programs like miRANDA (for instance many are referred in miRecords).
It can be, but at present all the examples we have written simply use miRBase targets.

> How is the  (miRNA-target_gene) prediction, from computational programs like miRANDA, accounted for by CORNA ?
We simply rely on other people (miRBase targets, microRNA.org etc) to do this and perform all the necessary quality checks

> How is the biologists' (miRNA-target_gene) validation  accounted for by CORNA ?
There are no specific tools, but if you want to load TarBase into CORNA, I am sure that would be not too difficult.

Your original question was this:

> I have to find humans miRNA sequences and the correspondent validated and/or predicted gene sequences.
> The only database I know of is  http://mirecords.umn.edu/miRecords where all such information is accessed through a set of structured web site  pages.
> I wonder whether R or C++  provides an interface to automatically
> extract such data from miRecords or any other similar database.

You can get a data.frame of miR to gene target using CORNA:

library(CORNA)
targets <- miRBase2df.fun(url="ftp://ftp.sanger.ac.uk/pub/mirbase/targets/v5/arch.v5.txt.homo_sapiens.zip", slim=false)

You can then use biomaRt to extract the actual sequences, if you need them.

Thanks
Mick

From: mauede at alice.it [mailto:mauede at alice.it] 
Sent: 18 June 2009 19:48
To: michael watson (IAH-C); Michal Okoniewski; Wacek Kusnierczyk
Cc: bioconductor at stat.math.ethz.ch
Subject: question about CORNA


I browsed through your paper.
It looks like CORNA is not another pair (miRNA - target gene) prediction program like miRANDA and so on ...  ?
If I am guessing right (am I ... ?) then what is CORNA basing its miRNA -target_gene affinity score on ?

In my mind I have the three phases that miRANDA uses to score such "affinity". Phase-1 implements some empirical Biology rules, phase-2 implements free energy minimization, phase-3 implements conservation of species. Each phase is a finer screening that assigns a score to the target match. The final score is the total.

I cannot understand which criteria (empirical or computational) are at the base of CORNA assessment.
Is it a synthesis of the predictions of the many programs like miRANDA (for instance many are referred in miRecords).
How is the  (miRNA-target_gene) prediction, from computational programs like miRANDA, accounted for by CORNA ?
How is the biologists' (miRNA-target_gene) validation  accounted for by CORNA ?
I read it uses a number of statistical tests but I cannot figure out which data the statistics is applied to ... maybe because
I am not familiar with the information stored in the databases mentioned in the paper.

Thank you.
Kind regards,
Maura


Alice Messenger ;-) chatti anche con gli amici di Windows Live Messenger e tutti i telefonini TIM!
Vai su http://maileservizi.alice.it/alice_messenger/index.html?pmk=footer