[EMBOSS] use water/matcher to find where RNA bybridizes

[pmr at ebi.ac.uk]

Guy Bottu writes:

> One of our users had a problem : how to find the location where a small
> molecule of RNA binds to a mRNA and so interferes with its functioning.
> Nothing in EMBOSS and nothing found on the WWW. We finally did the
> following : use revseq -nocomp to reverse the mRNA and then align the two
> sequences using as matrix :
> -------------------------------
>     A   T   G   C   S   W   R   Y   K   M   B   V   H   D   N   U
> A   0   5   0   0   0   5   5   0   0   5   0   5   5   5   5   0
> T   5   0   5   0   0   5   0   5   5   0   5   0   5   5   5   5


..........

> -------------------------------
> This gave a reasonable result. water made the following alignment :
> ------------------------------


.....

> mRNA            2892 AATGTTGTGTGAGGATAATAGTAATAGTAATAGTAATAATAATAATAAT
> 2940
>                      ..  . . .... ... .. .. .. .. ..  ................
> RNAi               1 TTTGACCCTGCTACTACTACTACTACTACTACGATTATTATTATTATTA  49
> --------------------------------
> The only thing which bothers me is that the base pairs (which do have a
> positive comparison score) are not labeled as "similar", they get a '.'
> instead of a ':'. Does someone know why this is ?


I believe this is simply because the bases are not identical. A user
matrix can have arbitrary values, so the results are marked as similar
(A=T scores 5) but identities are only scored at zero and so never appear
with ":".

You could try setting the scores to match the hydrogen bonds for this
experiment (G=C 3 A=T 2 G=T 1)

RNA folding is a missing area in EMBOSS. The Vienna package has been
suggested as a possible EMBASSY package. Does anyone have any experience
with it, or suggestions for alternative RNA packages we could use?

regards,

Peter