HiSeq1000 readcount differences

[shubho]

I have recently joined a sequencing facility, and came across a problem that seems to frequently creep up in their Illumina HiSeq1000 - Significant differences in readcounts for libraries loaded on the same lane. Sometimes, the differences are quite large (16M for replicate 1 and 31M for replicate 2, for example). Can anyone tell me if this is normal, and a probable reason for why this should happen?

Thanks.

[pmiguel]

It can be tricky getting even read counts for multiple samples in a pool. qPCR or fluorimetry can give you an estimate of the titre of a given library, but sometimes this estimate is off.
The most accurate estimate of a library titre comes from clustering the pool and seeing what you get. But that's a "catch-22".

We now generally put all the samples we can in a pool and sequence 1 lane. Then based upon the results, we re-pool and attempt to correct for any major discrepancies in the read count numbers per sample. Then we sequence subsequent lane(s) of the project with the corrected pool.

But this does require splitting a project across multiple flowcells. So it takes longer. And can only be used if the project needs at least 2 lanes of sequence.

[shubho]

Thanks pmiguel.