Mirna Mirdeep2 : Mirnas_Expressed_All_Samples.Csv Or Result.Csv Which One To Consider?

Question

[Rm]

I have a basic question: From the mirdeep2 outputs which one we need to consider for expression comparisions across groups (tumor vs. normal): (Or how researchers report?).

in the below example: mmu-let-7c-5p have different precursors...How to tackle this; I appreciate your inputs.

And also how does programs like "Deseq" handle differences in the initial number of reads for samples.

miRNAs_expressed_all_samples.csv

mmu-let-7c-5p   1452    mmu-let-7c-1    1452    1452    1452.00
mmu-let-7c-5p   1452    mmu-let-7c-2    1452    1452    1452.00

or result.csv.

chr16_6774      5.3     85 +/- 20%      -       729     729     0       0       yes     mmu-let-7c-5p   aae-let-7       -       -       ugagguaguagguuguaugguu  cuguacaaccuucuagcuuucc  ugagguaguagguuguaugguuuagaguuacacccugggaguuaacuguacaaccuucuagcuuucc     chr16:77599897..77600004:+
chr15_6481      0.3     12 +/- 12%      -       733     733     0       0       no      mmu-let-7c-5p   aae-let-7       -       -       ugagguaguagguuguaugguu  ccacaacauccagcucuacg    ccacaacauccagcucuacgccaagacugacugacggccuuuggggugagguaguagguuguaugguu    chr15:85536976..85537086:+

Answer 1

[Rm]

I got the answer from the mirdeep2 Authors:

"I would use the counts from the file miRNAs_expressed_all_samles.csv. These counts are based on the annotated mature sequence plus some small flanking sequence while the result.csv uses the estimated mature sequence by miRDeep2.

Regarding the normalization I think taking RPKM is a good idea and something similar is done in the qunatifier output. Since I am not an expert in differential expression analysis I can only recommend to you to use some package that implements these tests for example the R package DeSeq."

[Rm]

I have answered similar question seqanswers: http://seqanswers.com/forums/showthread.php?t=28976&highlight=miRNA