Hi everybody, I have some fastq reads from miRNA-seq and I want to classify them using the blast algorithm, but, in the miRdb all sequences are annotated in RNA nucleotides instead of DNA, I know that translation to DNA is not a complicated issue on linux command line but, if I do that, is the best way? Someone has had the same problem? And second, somebody knows if I will have problems looking for target genes using bowtie to align the reads? (due to align very short reads)
Thanks :)
•
written
by
Buffo
If you are trying to align miRNA reads to miRNA transcripts and figure out which miRNA is which, Bowtie is fine. However, it sounds like you are also trying to find target genes with Bowtie, and that will likely end in disaster.
miRNAs are very short, and their binding to mRNAs is famously messy, full of loops and mispairings. Bowtie is not designed for this task -- your false negative rate will be sky high. Also bear in bind that the entire mature miRNA need not bind to the mRNA, which will further challenge Bowtie and similar aligners.
There are many specialized miRNA binding-site prediction algorithms, for instance this list .
Practical Aspects of microRNA Target Prediction is a very useful paper for understanding this. The CLASH-seq paper is also very illuminating.
Thanks :), Actually I did that prediction, and I think it was not so bad, I get about 80% of aligned reads (with multi-hit). I will read that papers THANK YOU. Also there is a webserver for predict target genes from mir_id:
http://zmf.umm.uni-heidelberg.de/apps/zmf/mirwalk2/
I will use it for target prediction and anotate new ones after validation.
Log in to answer this question.
RNA should be recognized by BLAST.
Of cause you can convert RNA to DNA by an easy solution using
awk:awk '{ if (!/>/) { gsub(/u|U/, "T", $0) } print }':And if the query sequences are very short, use option
-task blastn-shortwhen usingblastn.