What is the specificity of a Microarray for human vs. mouse when completed for a mixed sample?

Question

[kruicci]

Hi there,

Can anyone comment on what the outcomes would be for using a tissue sample (human PDX tumour grown in a mouse) that likely contains both human and mouse cells for a microarray (to gauge RNA expression)?

That is, would mouse RNAs be able to hybridize? I am not familiar with the chip selection process for a microarray (I have never done one before) but I assume I would choose a chip whose probes were human sequences? I am interested in expression of RNAs from the human tumour cells collected after this PDX mouse model was treated with a drug.

I am not clear on how much similarity there would be between human and mouse RNAs and thus whether I need to worry about mouse RNAs binding to the (human) probes or not?

Of course, all of this could be avoiding by depleting the sample of mouse cells before proceeding with the array, but unfortunately that will not be possible for me at this time.

Looking for any insight in this area.

Many thanks!

[Devon Ryan]

I suspect that you could indeed run into some problems with cross-reactivity. Do you have any bioinfo folks locally that you could ask to check the probes on a couple chips that you're interested with? They can just map them to the mouse genome and see how many don't have an unreasonable number of mismatches.

Edit: I should add that you're unlikely to find a human microarray that won't pick up any significant mouse signal for at least a few probes (evolution and all), but being able to exclude probes prone to cross-contamination would prove good enough.

[russhh]

I would place no faith in any microarray experiment along these lines - especially if you can't immunodeplete the mouse cells. RNASeq is the only route available in this setting (even then, you have to be pretty careful)

Answer 1

[Juan Cordero]

Hi,

it might be a bit late answer, but I dealt with such problem a year ago. What I performed was, in summary, the exclusion of human microarray probes that aligned to the mouse transcriptome with <= 2 mismatches. I used Clariom D human hg38 arrays from Affymetrix. To flag probes to remove, I used blat with the following parameters:

• tileSize=12
• stepSize=7
• minScore=23
• oneOff=1
• repMatch=1000000

Check Methods from this paper.