Mirdeep2 Quantifier Module Output

Question

[liran0921]

Hi everyone,

I am using the mirdeep2 quantifier module the quantify the miRNA expression with the default parameters.

As some miRNA can be mapped to multiple locations with different counts number, I am confused at how to determine the real counts number:

eg:

#miRNA         read_count    precursor
bta-let-7a-3p    994    bta-let-7a-1
bta-let-7a-3p    994    bta-let-7a-3
bta-let-7a-5p    230845    bta-let-7a-3
bta-let-7a-5p    231026    bta-let-7a-2
bta-let-7a-5p    231087    bta-let-7a-1

we can see that bta-let-7a-5p is mapped three times with slightly different number, so how should i choose the counts number? I would like to make a reads counts matrix for all the libraries I sequenced, which need a unique number for each specific miRNA.

Thanks.

[Pam]

Hi liran0921 and others, I need some help in quantifier module of mirdeep2. I need to quantify some deep sequencing reads for differential expression analysis. I assume i should just go ahead with quantifier module with matrue , precursor and read sequences (as per the documentation)?!. Or should i map them to genome and then proceed with quantification?

P.s: the reads are already adapter trimmed.

Answer 1

[Martombo]

the problem is that the mature miRNA - precursor miRNA association is not one-to-one but many-to-many. that means that the same mature sequence can be encoded by different precursors and that one precursor has got more than one mature sequence. the information you get with a standard small RNA-seq experiment is usually almost exclusively about the mature sequence. the reason for this is that the mature sequence are the ones characterized by a much higher half-life in the cell (compared to the precursors). so if you sequence the small RNA you mainly get just the mature sequences, which is eventually produced by more than one precursor along the genome. still, mirdeep2 can try to distinguish to what extent each precursor is contributing to the mature miRNA. this is done by using a few base pairs unique to one precursor (just before or just after the mature sequence) that might be sequenced (see figure from miRBase: these histograms represents reads mapped to some miRNA with the same mature sequence. the pink columns represent the coverage of the mature sequence, while the blue ones are in this case unique to the precursor).

so, what I did with the mirdeep2 output was to compute the average for each mature miRNA (rounded up) and used it as raw counts for subsequent differential expression analyses. that would represent the average expression of a mature sequence from its precursors. (in no case the difference was really high)

[Martombo]

so just to be more clear: if you get a sequence that maps to the miRNA in the figure and ends with a UGG it means it's coming from precursor 1, if it ends with a UAG it means it's coming from precursor 2. the problem is that, as you can see, only few reads extend into the precursor, making it hard to guess the respective precursor contribution.

[liran0921]

Thanks for your reply. I am using the largest read count value when there are multiple mappings for a miRNA. I suppose this is not big difference as you said.

Answer 2

[liz9090950]

could you tell me what the read count means. does it means expression quantity? thanks a lot

[liran0921]

yes, the read count can represent expression quantity.

Answer 3

[meetgirish1986]

Will it be appropriate to consider the average of the read counts for each mature miRNA in the following scenario?

#miRNA         read_count    precursor
bta-let-7a-3p    0    bta-let-7a-1
bta-let-7a-3p    3    bta-let-7a-3
bta-let-7a-5p    2    bta-let-7a-3
bta-let-7a-5p    0    bta-let-7a-2
bta-let-7a-5p    0    bta-let-7a-1

Thanks

[eager_learner]

I think in this scenario it would not be advisable as they are different miRNAs (5p and 3p), though coming from the same precursor but different arm.