RNA-seq analysis- finding isomiRs in BAM files?

[andra]

I am dealing with seq data for the first time and currently I am analyzing some Illumina RNA-seq libraries for identifying isoMiRs.
I have processed the raw data using different tools (FASTX,FASTQC,GALAXY) and I am stuck at the mapping step.
I have aligned my reads to the pri-miRNA human sequences (from MirBase) using Bowtie and I have converted the obtained sam file to a bam file which I've planned to analyze using RSamtools...However I am not sure if that is the best option...


What is the "standard" procedure for analyzing the alignment files for dentifying isoMiRs?

Thanks.

[moriah]

Hi,

I am not sure about the "standard" procedure bur there are few steps you should take before any analysis of micorRNA
1. You mentioned that you are using RNA-seq library is it small RNA-seq or the total transcriptome? In the second case you can't analyze microRNAs from this data.
2. How long are your reads? The standard read length of Illumina is 35-38nt or 50nt. As you might notice none of these lengths resemble average microRNA length. In this case you have to trim the adapters that were ligated to the mature microRNAs and sequenced with them as well.

Regarding the mapping : If you are using Bowtie you have to carefully examine the parameters that you use, take a look at the bowtie manual and decide for yourself how many mismatches you want to allow (since you are looking for Isomirs you might want to be somewhat flexible with this parameter) other parameters that you might want to carefully use and adjust -m, --best,--strata.

If I remember correctly there are no pri-mir sequences in mirBase but pre-mirs (hairpin loop).

I am not sure what kind of analysis you want to run with Samtools on the data when looking for isomirs, but I do strongly suggest that you use the IGV viewer to look at the *.bam files before running any analyses so that you will get an idea on how your data looks like...

Good Luck!

[andra]

Hi moriah and thanks for the reply.

1. It is only small RNAs.

2.I have trimmed the adapters and discarded the the reads lower than 18 nts and the ones which had low quality scores. Also I have trimmed the reads and now I am using reads with an average length of 22 nts.

Regarding the mirBase sequences, I am using the pre-mirs from mirBase.

I will try to have a look at the data using IGV viewer...Otherwise I am not really sure how I can proceed after generating the (assumingly right) alignment file . Do you have any suggestions?

[moriah]

I am no expert with isomirs identification but there are some tools the you might want to get familiar with, like BedTools (coverageBed might be useful). Since the isomirs are defined as variants of the mature microRNA there also align differently to the reference with small shifts from the mature miR coordinates of they fail to align to few nucleotides on the reference seq downstream to the mature seq usually with A's or U's (in rare cases upstream to mature mir as well) , coverageBed tool intersects between your reads and a bed file which contains the coordinates that you wish to intersect with (you can use the mature mirna coordinates). For the reads that will deviate from matures you can check for their abundance to claim for an Isomir pattern. Regarding the Isomirs that went through ADAR modifications you can try to use the "pileup" Samtools function which supposed to find SNPs but might be useful in this case although I am not sure if it is the best way.

There might be a more accurate and perhaps easy pipelines but for the sake of brainstorming I tried to suggest possible directions...

Good Luck

[andra]

There might be, but I think that at this point this is very valuable information..I haven't found anywhere documented the whole procedure..The rna-seq analysis descriptions usually jump from the alignment straight to the findings... Thank you.