Why does number of reads per sample vary?

[MD19]

Hi all,

I am just starting to learn NGS data analysis.So this may be a very basic question. So in the data for your samples which were normalized in quantity for sequencing run which are expected to have say 10 million reads each, what does it mean if you get a Gaussian distribution averaging at 10million but with +/- 3 million reads? how do you explain that?

Thanks

[Bukowski]

Well - you try pipetting so accurately that you can manage better...

Quantification isn't exact. Dilution isn't exact. Pooling isn't exact.

[MD19]

Thanks, Bukowski! Could it also be 'quality of the sample' issue affecting the final output? Just trying to think of all possibilities in this scenario.

Well - you try pipetting so accurately that you can manage better...

Quantification isn't exact. Dilution isn't exact. Pooling isn't exact.

[SNPsaurus]

Sure, fragmented DNA may make libraries that have a shorter fragment length. Unless you are precise in figuring out how many fragments are in your sample's library, the normalization will be off (and it is a given that you won't be precise at this since fragments can tangle and daisy chain which affects the apparent size distribution. If they are ever pooled and amplified, the short fragments will be enriched in the pool as well.