Hi all,
I am currently using data obtained from the HumanOmni2.5-8v1_A chip to call CNVs in my samples. The project contains 80 samples (74 cases, 5 controls [one being a duplicate of itself] and 1 HapMap sample ran by the facility that processed our samples). My question is in regards to clustering our data. We decided it was best to perform two separate analyses, one in which clustered based on our samples and one in which we clustered using Illumina cluster files (created from HapMap samples run on this chip). We felt that if we only cluster based on our data, it could affect CNV calling for CNVs that are present in a large proportion of our project (remember 74/80 are cases).
However, when clustering using the standard cluster file we see fairly poor call rate (>50% of our samples with less than 98% call rate). I am using Illumina's tech note "DNA analysis - Infinium genotyping data analysis" to evaluate the clusters and see about reclustering some/zeroing as necessary. Even clusters that look like they have distinct groups are showing a large variance in Norm R... I'm just wondering if anyone has any experience with using the standard cluster file vs. clustering based on your own data.
Thanks.
-Bob
