help in my miRNA analysis

Question

[Claire]

Hi Everyone

I am trying to detect miRNA in my miRNA samples. Just want to check I am following correct steps:

1. I downloaded mature.fa from miRBase which is the fasta of all miRNA published and annotated.
2. I indexed this mature.fa using bowtie
3. I trimmed my reads and aligned them with bowtie to the mature.fa (using the index I build)
4. Percentage of aligned is 0%, this means no miRNA is found?!

Can anyone help me if there is any step missing?

Thank you

[cpad0112]

check your reference fasta file

Answer 1

[colindaven]

• try fastqc before and after trimming
• If adapters are still present in the reads after trimming, trim again
• repeat above
• maybe try a different trimmer too
• a genomic alignment is also useful (not just to mature miRNA
• bowtie1 cannot align indels, so bowtie2 might be more helpful if some reads do contain indels.

Possibly also try a specialist miRNA analysis tool

[Claire]

Ok, thanks so much. Will check trimming and try bowtie2. I tried miRDeep2 but have issues with it, please let me know if you suggest another tool. Thanks a lot. But if I align to the genome how I get the miRNA alignments? Thanks

[colindaven]

I haven't used this one personally, but the nf-core pipelines are normally excellent community analysis pipelines.

https://nf-co.re/smrnaseq/output

[Claire]

thank you :)

Answer 2

[Apex92]

Maybe you have not converted the Us in the mature.fa to Ts that is why you do not have any mappings.

Generally, I would say first try to know the structure of library preparation, and then with knowing your adapter sequence you can trim them using cutadapt.

If you have UMI in your library structure then correcting for UMIs is necessary.

Then when you have your reads UMI corrected (if UMIs are there) and trimmed, you can map them using quantifier.pl from a well-known tool called miRDeep2 and this will give you a count matrix of miRNAs in each sample (there is also a tool called mapper.pl from miRDeep2 as well if you only want to map - just check the link I shared).

To be able to use quantifier.pl you need to provide both precursor and mature sequences. An example command line for quantifier is:

quantifier.pl -p hsa_hairpin_miRBase.fa -m hsa_mature_miRBase.fa -r reads.fa -t hsa

The reads.fa is a concatenated file from all your reads. Just be aware that the header of your reads after the > sign should be your sample ids but not more than 3 words otherwise the quantifier.pl will give an error (for example >SA1).

I hope these are helpful for you.

[Claire]

Thanks so much Apex92. I will try this, thanks.

Answer 3

[Carlo Yague]

0% mapping is weird, there is probably an issue with your reads. With small RNA-seq, read length often exceed miRNA length so usually, the reads contain part of the adaptor sequence which might interefere with the mapping. I see that your reads are already trimmed, but have you checked read quality and possible remaining adaptor contamination with FASQC ? If not, my advice is to start looking in that direction.

[Claire]

Oh, thanks so much. I will double check that.