PhiX SNPs and base recalibration?

[dsobral]

Hi,

I was thinking of using PhiX library sequences from a run of our MiSeq to build a base calling error model and recalibrate our read qualities for other samples. First I would like to know if this is a good approach (shouldn't the MiSeq software do something like this anyway)?

Then I was surprised to see that even that library has 3 SNPs relative to the "official" reference NC_001422:
phix 1301 . A G 3070 PASS DP=48041 GT:GQ 0/1:3070 0/1:3070
phix 2793 . T C 3070 PASS DP=48247 GT:GQ 1/1:3070 1/1:3070
phix 3133 . C T 3070 PASS DP=47638 GT:GQ 1/1:3070 1/1:3070

Is this normal? Is everyone observing the same SNPs or does it change?
Is it sequencing error or real differences in the library provided by Illumina versus the reference (I'm assuming they are real differences)?

Thanks,
Daniel

[microgirl123]

I'm too busy to look up a phiX test run, but, yes, there are a couple of known SNPs when you compare Illumina's phiX to the reference genome.