microRNA- sequencing analysis

Question

[kanwarjag]

I have a microRNA seq data (RNA seq after depletion of rRNA) using Illumina TruSeq Small RNA kit. Once I have the FASTq files as out put shoudl I be aligning to genome (suh as hg19) and than map the miRNA from a bed file from any mirRNA db or alternatively just align to mirRNA db.IS there an speific aligner to align to miRNA db or only blast is the option. What should be the best/ traditional approach.

Thanks

Answer 1

[David Langenberger]

Simple strategy:

1. Clip the adapter sequences at the reads 3' ends
2. Map your reads against hg19 using a NGS mapper, like Bowtie2, BWA, segemehl, STAR, etc.
3. Download the microRNA annotations for hg19 from miRBase
4. Do you downstream analysis (e.g. quantify your microRNAs using e.g. featureCounts)

[kanwarjag]

Thanks David, One clarification do we trim reads keeping in view small RNAseq before aligning ? will it of advantage or alternatively just do the quality control and move with alignments.

[David Langenberger]

Very good point! You actually have to clip the adapter sequence before you map the reads! Sorry, that I forgot that.... haven't done small RNA-Seq analyses for some time now.

[kanwarjag]

David, Do you know if I can get bed/ GTF file for GRCH37 annotation from mirBase . I am using hg19 and I see the current gff3 file is of GRCH38..
Thanks

[David Langenberger]

Go to http://www.mirbase.org/ftp.shtml and click on 'Previous releases'

[David Langenberger]

Or use LiftOver. Explained here

Or go to the UCSC Table Browser and download the microRNA annotations of the assembly you need (track: 'sno/miRNA').

[apelin20]

Smallest Illumina sequencing is 50bp as far as I know. Your microRNAs are going to be ~21-23nt. If you do not trim, half of your read will be adapter sequence and will prevent proper mapping. One question I do have however, is for step 4. I use bedtools multicov to see how many reads overlap with mature miRNA annotations. I ask for the read to be of the same strand as the annotation, as well as 100% overlap between the read and the annotation. Can anyone comment on this?

[David Langenberger]

On the same strand is important, since some microRNAs are known to have an anti-sense gene. 100% overlap of the read with the precursor and vice-versa? Or only 100% of the read? Or do you distinguish the 3' and the 5' mature sequence? I would go for something like 80%, since the annotations of the mature sequences are only for the most dominant mature sequence, but there are also shifted ones.

Also allow for errors in the alignment (RNA-editing events within the mature sequence are known) and turn on the detection of multiple mappings loci in the mapping tool (microRNAs, especially mature sequences, are known to occur in multiple copies in the genome).

[Arindam Ghosh]

Can we use annotations from Ensembl in step 3 as it contains co-ordinates from miRBase?