Differences in trimming between Novaseq and Nextseq outputs

[peterashtonEMHP]

Hi

We are running the same sequencing library (constructed with DNAprep) on two different Illumina machines, and are seeing differences in trimming with Trimmomatic. The output from the Novaseq loses around 30% of read pairs (when set to only retain pairs where both reads are retained after trimming), compared with only 5% being lost from the Nextseq. Fastqc quality plots and read length profile look identical for the two machines, and we can't see any other difference (other than the change Illumina made to the quality scores - but these should make the Novaseq data look higher quality than the Nextseq).

Any ideas why this is happening?

Thanks
Pete