Replication Of Gwas Hits From Illumina Quad Platform Using Affymetrix 6.0 Platform

Question

[Khader Shameer]

I have a bunch of dbSNP IDs (SNPs from a GWAS after discovery and replication). I need to replicate this SNPs from Illumina platform on additional cohorts. One of the cohort genotyped the data using Affymetrix 6.0. When I mapped the dbSNP rs#s using GenomeWideSNP_6.na24.annot.csv, I noted that 25/98 were only present in the platform. I am wondering why such a difference between the platforms ? Is this expected ? What would be an ideal way now to replicate my Illumina based SNPs in this Affy dataset ? Imputation, LD based search or any other suggestions ? I am looking for suggestions to solve this issue.

Sample IDs that are not mapped to Affymetrix 6.0 
rs1011663    
rs10277544    
rs10495490    
rs10503511    
rs10512401    
rs1073068    
rs10765931    
rs10889394

Answer 1

[QVINTVS_FABIVS_MAXIMVS]

The companies use different probes. I've worked on a consortium array project on multiple chips with multiple genome builds. It's quite a headache. But you only need the union of probes if you wish to use both platforms in your experiment. If you like to only replicate the experiment on the Affymetrix platform you can use all the Affy probes.

For LD, you expect the same regions to be in LD with each other. That is to say the same genomic positions should be in LD, but the SNPs in those regions will be different across platforms.

If you want to include SNPs from both studies you need the union of the two. However, as you noticed your resolution will be compromised. Also I'm not sure a normalization step is necessary for genotyping, I worked on the LRRs (which need to be nornalized by plate) for CNVs

You can increase your resolution by imputation, as you suggested.

TL;DR: If you do not need the Ilumina samples you can perform the same experiment without removing affymetrix probes. To do comparison you need the union of probes.