Combining high throughput data from Illumina

Question

[maggiepan]

Hi All,

I wonder if I could combine data generated by different Hiseq platforms, say 2500 vs 4000?

Thanks!

[WouterDeCoster]

What does "combine" mean in your analysis? Please be as specific as possible.

For example: If you are doing differential expression analysis with treatment A on HiSeq2500 and controls on HiSeq4000 you are potentially introducing a batch effect.

[maggiepan]

precisely, by 'to combine' I meant to do DE analysis for both set of samples. and yes you are right! it is actually a batch effect question, i would have thought of that, thank you! do I have to do correlation analysis first to check at least whether I can combine the two set of samples for analysis?

Answer 1

[Devon Ryan]

Sure, though keep in mind that if the library prep was done at different times/by different people/etc. that you have a batch effect to account for.

[maggiepan]

That is very helpful, thank you!!