siRNA target selection problem - (Jan/20/2003 )

We just begin working with SiRNA in my lab. So I'm searching the sequences for this SiRNA. I know that they have to be specific...but when i blast the potentials sequences, I always find an another similar sequence not related to my target. In the best case, the potential sequence bind only 17 bases (the sequence lenght is 21 bases) of an unrelated sequence from mouse or human (I work with mouses)...is it enought specific for an SiRNA inhibition experiment?

thanks in advance for you responses

A smilarity of 17 out 21 is significant. Such a siRNA may cause off-target effects. Are those unrelated targets mRNA seqeunces? If not, it may be OK.

Is there any method to target selection? Do most people look through gene for AA?
If anyone can suggest good papers for designing that would be great.
Cheers,
cat

Target selection is not so challenging as it sounds, though there are many rules and algorithms.

Here are some resources on the web:

The original Tuschl's rules:
http://www.rockefeller.edu/labheads/tuschl/sirna.html

Rational siRNA design paper:
http://www.ncbi.nlm.nih.gov/entrez/query.f...t_uids=14758366

This nice Excel macro implements the algorithms reported in the above paper. I have used it for designing all my siRNAs. I recommend it.
http://ihome.ust.hk/~bokcmho/siRNA/siRNA.html

Thanks for that info pcrman - it's been a great help.

Can I just clarify - once you have used the Excel program you blast the DNA sequences, then any good candidates you use the RNA seq to order siRNA? Do you add the Tuschl TT? (if so where/why - I've never quite got my head around this).

Also, how many synthetic siRNA do you normally begin with to check function?

Thanks again,
cat

Yes it is bit confusing at the begining.

The excel program gives a list of targets and corresponding siRNAs ranked by total score. the siRNAs are 19 nt. First, you have to do a blast [using not just the 19 bp but incluing some more bps both upstream and downstream because UCSC Blat program (my favorite) won't accept queries that are too short]. If there is no obvious homology with other non target sequence, I will pick up those with high scores. At this time you can take Tuschl's rule into consideration such the target should be xx bp downstream of the ATG and ect. After you find a desired target, usually you just copy the siRNA sequence (19 nt) into the vendor's order form, if you choose add dTT to the 3 overhang, they will be added automatically.

For example:

The algorithms reported in the Rational siRNA design paper:
http://www.ncbi.nlm.nih.gov/entrez/query.f...t_uids=14758366

is also implemented in the following program (SVM RNAi):

http://www.changbioscience.com/stat/sirna.html

You would need to download the stand-alone version, which is currently 3.6. The demo is fully functional.

If your coding sequences keep coming up with too many non-target sequences, a fun thing to do is to make a RNA hairpin corresponding to the 5'UTR instead. This should silence your gene just as well and may lead to less non-target attacks.

yes you are right.most people look through gene for AA and another 19 nts.Also it depends on the vector you used (H1 promoter or sp6).Do not worry,usually it will work,the difference is the effience.