How to align reads to known ncRNA database

[skblazer]

I want to align a lot of reads to the known ncRNA database.

I don't know which mapper is fit for small RNA alignment.

blast? megablast? or bowtie ...

how to set the parameter?

Thanks

[colindaven]

You'll need to mention what read length and system you used for the sequencing. Generally BLAST is too slow for "a lot of reads".

[dcfargo]

DSAP aligns to both miRBase (16) and RFam (10) if you have very short read data

http://dsap.cgu.edu.tw/

[skblazer]

Thanks

My question is if I used blast/megablast, I'll get a lot of alignments without perfect aligning rate. How to set the cutoff to filter the output?

For example, if one read's 6-26bp (27bp in total) can be aligned to a known tRNA, is it a candidate tRNA or not?

For tRNA or rRNA, I should use a more relax alignment than miRNA, am I right?

[bioinfosm]

another option could be to align all reads to the genome without biasing to a selective reference dataset, and then use coordinates of your ncRNA database of interest, and identify which ones are expressed?

[skblazer]

Because the organism I'm working on does not have a reference, I have to align them to some ncRNA database.

I don't know the if the similarity of tRNA or rRNA among the organims is high as well as miRNA.

another option could be to align all reads to the genome without biasing to a selective reference dataset, and then use coordinates of your ncRNA database of interest, and identify which ones are expressed?

[colindaven]

To the best of my knowledge rRNA and tRNAs are _more_ conserved than miRNAs. At least in bacteria we work on there is considerable variation in miRNAs between closely related species.
I'm sure you can find some good references on this in pubmed though.

http://www.biomedcentral.com/1471-2164/8/481

"Our results suggest that while there is a conserved set of miRNAs among plant species, a large fraction of miRNAs vary among species"

Ultimately you'll need to consider topics like thermodynamic stability, perhaps this link is helpful ?

http://www.biomedcentral.com/1471-2148/10/329/abstract

[sdarko]

Thanks

My question is if I used blast/megablast, I'll get a lot of alignments without perfect aligning rate. How to set the cutoff to filter the output?

For example, if one read's 6-26bp (27bp in total) can be aligned to a known tRNA, is it a candidate tRNA or not?

For tRNA or rRNA, I should use a more relax alignment than miRNA, am I right?

I've been using BLASTn to align snRNA reads to tRNA, rRNA, snoRNA, piwi associated RNA and miRNA.

Basically, I first trim 3' adapter sequence and collapse the reads. Then I use BLASTn and use '-perc_identity 100' and '-word_size 16' so that at least 16 bases have to perfectly align to get a hit. When I'm parsing through my results I compare the length of the transcript and the length of the alignment. If they're the same, I call it a good alignment. If they're not, I set it aside to align to my next reference set.

[skblazer]

Thanks for your advise.

I'll try.