Dear Bioconductor Community,

I want to understand the function "dropLociWithSnps" within minfi

GRset <- dropLociWithSnps(GRset, snps=c("SBE","CpG"), maf=0)

minfi developers said

Probe, CpG and SBE correspond the SNPs present inside the probe body, at the CpG interrogation and at the single nucleotide extension respectively. We strongly recommend to drop the probes that contain either a SNP at the CpG interrogation or at the single nucleotide extension.

Does anyone know the meaning of CpG interrogation and single nucleotide extension(SBE)?

There is no explanation about them.

I think these words are key words which help me to understand why I need to drop the probes that contain either a SNP at the CpG interrogation or SBE.

Thank you in advance.

Best,

Yujin Kim.

There are two types of probes used by Illumina. The type I probes bind directly to the CpG site and have two different probes (one for a methylated site and one for an unmethylated site). The type II probes are different; there is just one probe for a given CpG that binds directly adjacent to the CpG site. The status of the site is interrogated by performing a single base extension. See here for more information.

You want to drop probes that have SNPs for two reasons. The first is sort of obvious. Treatment with bisulphite converts unmethylated cytosines to thymidines, so a GC site becomes GT. If there is a C/T SNP at that locus, you cannot say if a subject with GT at a given locus is due to the SNP or due to methylation. The second is less obvious. Both probe types are 50mer probes, and depending on the GC content of the binding site, the binding of the probe may be reduced due to mismatches in the sequence due to the presence of one or more SNPs in the binding site. This is particularly true for SNPs that are close to the CpG site.

The three types of SNPs that are provided by minfi are those with SNPs in the probe binding site, SNPs that are in the binding site for the CpG (for type I probes), and SNPs that are in the SBE (for type II probes). The last two are more problematic than the first one, so you can decide which you want to drop. The authors state

We strongly recommend to drop the probes that contain either a SNP at the CpG interrogation or at the single nucleotide extension. The function dropLociWithSnps allows to drop the corresponding probes. Here is an example where we drop the probes containing a SNP at the CpG interrogation and/or at the single nucleotide extension, for any minor allele frequency: