bias in merging sequence data?

[quinne5]

Hi,

Could anyone offer any advice as to whether it would create bias downstream in RNA seq DE analysis to combine reads from 2 different sequencing runs carried out on the same library for the same sample that are of varying sizes..e.g. one containing 3 million reads and one containing 30 million?
In the past I've combined fastq files of similar sizes(e.g.each containing ~20 million reads) before aligning but in this case I was wondering if things might be a bit off ...

thanks for any help!