Ethnic groups information from SNP arrays.

Question

[matthew.oberley]

Is it possible to derive information about ethnic groups from chromosomal microarray data (Oncoscan)?

Answer 1

[Kevin Blighe]

Good question.

It should technically be possible provided you have ethnicity-informative copy number variant (CNV) markers. However, using Oncoscan, I'm not confident that the predictive model would have high sensitivity / specificity because Oncoscan is tailoured for cancer. Using genotype data, it's certainly possibly to build highly predictive models, as to which I allude in these threads:

• Produce PCA bi-plot for 1000 Genomes Phase III - Version 2 ( formerly: Produce PCA bi-plot for 1000 Genomes Phase III in VCF format (old) )
• A: How to predict individual ethnicity information by using hapmap data

Coincidentally, years ago, in a rudimentary study, I defined 'haplotype-tagging CNVs' from the Internatiional HapMap 270 data (see Haplotype Classification Using Copy Number Variation and Principal Components Analysis ), which was able to segregate the HapMap populations very well via hierarchical clustering. These were defined from the copy number markers on the Affymetrix SNP 6.0 microarray:

If you were somehow able to merge your Oncoscan data with the data that I used, you could cluster everything together and loosely infer ethnicity by observing the clusters in which your samples sat. I am sure that some markers on both array types overlap.

Either that or you could build a regression model from the absolute copy number from the data that I used, and then use it to predict the ethnicity on the absolute Oncoscan copy number. This is all experimental, though.

Kevin