Regarding miRNA analysis

[deepika123]

hello to everyone,

Recently, i am analyzing sRNA data and i am trying different-different tools for this such as miRDeep2, PIPmiR and SiRNA tool kit.
Any one would suggest me that which tool will good for sRNA analysis and one more thing which tool will good for mapping of sRNA reads. Most of people used bowtie for this. I also used bowtie for mapping but when i do mapping sRNA reads against mature sequences of miRNA of different species then only .21 or .3% of reads are mapped but when in these tools more than 80 or 90% reads are mapped. I did not understand it. Please also suggest me that which parameter could be taken for mapping?

Please guide me for this analysis


Thanking You

[Brian Bushnell]

It's better to map short sequences against longer references than vice-versa.

[diego diaz]

Try this parameters with bowtie:

-n 0 -l 15

"-n" is for the number of mismatches in the seed and "-l" is the seed length. In other words, the first 15 bases can have 0 mismatches.

if you want to perform differential expression analysis, maybe you should use this parameter:

-m 1

"-m" discards all the reads with more than INT (in this case 1) valid alignments. (it is something like unique mapping)

bowtie 1 is more suitable for short reads than bowtie 2. I recommend you that you use all the genome as reference instead of miRNAs only.

[deepika123]

Thank You...for yours suggestion. One question is also in my mind when we align short reads against mature miRNA then only small number of percentage is mapped (using bowtie) while others tools such as mirdeep and other tools give high number of alignment percentage. how is it possible ???

[Brian Bushnell]

Well... different algorithms are always going to give different answers. You need to use the one best suited to the job.

If you are interested in matching reads to a set of miRNAs, it's possible that kmer-matching would be a better approach than alignment, but that depends on the goal of your experiment. Or, as Diego mentioned, mapping to the whole genome.

[aggp11]

Like Diego and Brian mentioned, you should try longer reference sequences like the whole genome (if available for your species of interest) or the miRNA hairpin sequences (can be downloaded from mirbase).