Do mice from the same strain have identical genomes?

[feralBiologist]

I am, actually, a bioinformatician, not a biologist hence my question may appear very naive. Here is my case:

I was given exome data of several transgenic mice. Some have been additionally treated to develop certain types of tumors, a single transgenic mice has been kept as control. I have been asked to look for tumor genome variation.

Most studies on tumor genome variation I know of use paired samples (a sample from the tumor and a control from a healthy tissue of the *same* animal or individual). This is still not trivial as the tumor is often a mix of different subclonal populations but, at least, there are plenty of bioinformatics tools that try to elucidate the tumor mutations vs the healthy tissue and all the differences what are found are atributed to the tumor.

What I don't know is whether I can treat the different transgenic mice from the same strain as having the *same* genome, i.e. attribute any differences between genome of the tumor of animal A and the genome of healthy animal B to the tumor alone, rather than to genome variation between A and B?

[dpryan]

That's a surprisingly tricky question to answer. The really short answer is, "No they're not completely identical", but a more useful answer is rather longer.

Mice purchased (or otherwise sourced) from a single vendor at a single time point will be very close to identical (some of the strains have known polymorphisms, which at least some of the vendors quality control when they breed). At least some vendors will replenish a line from a cryopreserved stock every 20 or so generations, so, depending on the vendor's breeding scheme (it's unlikely that a vendor creates each generation from a single breeding pair, so variation will tend to increase over generations), what generation of breeding you happen to get will affect how variable the genetics are.

BTW, most molecular biologists (if that's who you're working with) are pretty bad at keeping track of mouse substrains (those of us who also do behavior pay more attention to this). So, if they happen to have mixed, say, C57Bl6 mice from both Jackson and Charles River, then you'll have more genetic variability than otherwise and they're unlikely to have noted this.

Variability will also change depending upon how the transgenic mice were created. Usually, these are created in one strain and then crossed into another (often some variant of C57Bl6), so the amount of backcrossing (see the wikipedia article if you're not familiar with that) and how it was done will also affect things.

So, yeah, treating a single mouse as control will increase the false positive hits (as you rightly seem to fear). I can think of ways to ameliorate this somewhat, but since sequencing is pretty cheap, that might be the best route (assuming there are still mice left). Having said that, you might discuss with the biologists what their end goals are with the datasets. I can think of some goals that will be less affected by false positives than others, so things could be less-bad than feared.

Don't you love it when you're consulted after the fact?

[feralBiologist]

Thank you dpryan - this is a very helpful reply.

In this case it is, essentially, an exploratative study which aims to establish a list of tumor-related genome variations. Which is why I would like to avoid as much as possible any normal biological variation between cases and controls. I will ask whether they have any healthy tissue from the tumor-afflicted mice which might be sequenced and used as a control.

Yes, it is frustrating to be consulted after the fact - which seems to be more often the case than not. Although in this particular case the study has started before I joined the institute so I cannot blame anyone.