Did low input library and traditional library generated RNAseq data can be compared?

[bioinforD]

Hello,everyone
If there is sampleA using low input library to generate RNA-seq data
the other one sampleB using traditional library
Do they two sample(sampleA and sampleB) can be use to find different expression gene or other transcriptome analysis?
Does the method of library make a bioinformatic analysis bias?
Thanks!

[dpryan]

I would be rather hesitant in comparing those, particularly if your groups are partitioned by library creation strategy. You'd be best off adding library type as a factor in your model to try and compensate for any effect.

[bioinforD]

Thanks for your suggestion,but how can i do with this samples ,
usually,i use tophat to do mapping .

One sampe RNA only 100ng, so there no choice but low input library,the other can use traditional mehtod.

[dpryan]

If one group only has enough RNA for a low input kit, then you should use the same kit for the comparison group (ideally, use a similar amount of input). Otherwise, the output DE genes will be due to both the group-effect and a batch-effect.

[bioinforD]

thanks, you are right,maybe i should think about use the same low-input library for both samples.

[dpryan]

Yeah, as a general premise, you'll always be well served by limiting uninteresting differences as much as possible. If you can't limit them, then you have to control for them, which isn't always possible (or practical).

[TonyBrooks]

Thanks for your suggestion,but how can i do with this samples ,
usually,i use tophat to do mapping .

One sampe RNA only 100ng, so there no choice but low input library,the other can use traditional mehtod.

You are aware that most current RNA-Seq protocols can easily cope with 100ng of total RNA? We regularly use the Illumina TruSeq RNA v2 protocol on 100ng and produce libraries in the 50nM range, even when dropping PCR cycle number down to 12 cycles. We've also produced library from 10ng total RNA using the NEB Ultra protocol