Take the 2-minute tour ×
Biology Stack Exchange is a question and answer site for biology researchers, academics, and students. It's 100% free, no registration required.

I am trying to predict the secondary structure of certain predicted pre-miRNAs through mFOLD which is the generally accepted technique for structure prediction in most studies. I am finding it hard to interpret the result and accept if the pre-miRNA which I have picked and the miRNA I have can be accepted and included in my result.

This is one predicted pre-miRNA structure with the predicted miRNA position shown in green. enter image description here

Is this a good/bad structure and if so, can anyone give me a reason why. Thank you.

edit: I understand that energy levels are also something that is looked for in the prediction. In this particular figure, ΔG = -75.00 kcal/mol which I think is fine.

Sequence used for folding is

AAAAATAAAAAACAGAACCATGAGGCAGCGCACCAAGAGCGTCGTGAACGTCCGCGAATA CTTCCGCATGGACGGCAGTGGCGCCCCCGAGCAGGAGGACGACGATGACAACGATGACTG GATGGCCATCGCCATGCAGGGCGCACCGCGTAAGGTCAGCGTGGAAGTCGTTAAGCCTGG CAAGAAGGCACGCGCTCGTACGTCGACGTGTGACTCAC

and the highlighted match area is

ACGATGACAACGATGAC

So this structure should be fine because the sequence lies along a stem-loop structure. enter image description here

I am worried about that bulge on the left top corner though. That doesn't affect the predictability of this structure does it?

share|improve this question
    
As a non-specialist reader, I wonder when you say mMFOLD is a "technique" if you mean it's software? –  daniel May 26 at 14:44
    
@daniel yup software –  The Last Word May 26 at 20:30

1 Answer 1

up vote 1 down vote accepted

The green region will definitely not make an miRNA: It is not a part of a stem loop. See typical miRNA structures from miRbase.

EDIT

Bulges can sometimes determine which strand is chosen as mature miRNA. However, this green region is also is unlikely to form miRNA because the stem is just 15bp.

share|improve this answer
    
Please look at the edit of the question. –  The Last Word May 27 at 10:20
    
I didn't know if I should ask this as a separate question so, I am posting it as a comment. In one paper (bioinformatics.oxfordjournals.org/content/21/18/3610.abstract, I found that ~70 nt of the matched flanking area was taken as pre-miRNA while in this paper (hindawi.com/journals/ijg/2012/957607) ~100 flanking nt of the match area was taken for pre-miRNA structure prediction. Any idea why so? –  The Last Word May 31 at 7:29
    
no idea.. it is best to take 60-70 nt from either side (because you don't yet know that your read is -3p or -5p). I suggest that you use a machine learning algorithm to predict the right miRNA structure. Co-incidentally a friend of mine was working on similar problem and was talking to me about it (for a moment I thought you and her are same people :P ). I'll let you know if we make some progress. –  WYSIWYG May 31 at 8:11
    
I'll post it as a question then to see if anybody else could get some light on the issue.. –  The Last Word May 31 at 9:04

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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