theta value and R value in SNP data

Question

[ftp]

Hello

I have a question regarding SNP data. I have a set of SNP files belonging to treated and control conditions. These SNP data contain different attributes (theta-value, R-value, Log R Ratio). I'm having difficulty in understanding what each of these attributes means. If anyone could explain or knows a good tutorial to explain the meaning and differences between them I'd really appreciate it.

[karl.stamm]

I'm afraid basic statistics is not bioinformatics. Where did you get the data? Only the file generator knows for sure what theta and R mean, but it sounds like minor allele frequency and some correlation. A log ratio sounds like an effect likelihood. Maybe you want those to be high? Plot them as a function of position, and post a picture please.

Answer 1

[Irsan]

It seems like you have data from illumina snp arrays where the results are exported from genomestudio. Theta value is the same as b allele frequency. it ranges from 0 to 1 and represents the fraction of bases that are genotyped as the b allele (variant allele). 0 means homozygous reference (AA), 0.5 means heterozygous (AB) and 1 means homozygous variant (BB). The R value represents the fluoresence intensity of that probe of that sample. The log R ratio was obtained by dividing the R value of your sample by a baseline (maybe your matched control) and represents the ploidy (copy number) of your sample at that genomic position. LRR of 0 means copy number neutral, positive values mean copy number gains, negative values mean copy number losses

[curious]

I struggled with finding the mathematical explanation behind this transformation, so hopefully this helps someone. It is explained in the second paragraph of the background of this paper:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2572624/

The way I understood it is that for illumina infinium arrays the two channels:

Cy5 = Red = A allele = X signal Cy3 = Green = B allele = Y signal

The raw data is analyzed like this in GenomeStudio

1. The raw X and Y signal from each allele is normalized to account for background signal, etc using a proprietary illumina algorithm

2. These normalized X and Y signals for each sample can be plotted on a cartesian coordinate system to get this type of plot in GenomeStudio:

1. More commonly, the "polar transformation" of these values are shown with R plotted against theta like this:

https://imgur.com/dbRjBBp

• R is the intensities. R for a sample is the sum of normalized X and Y (R= X+ Y)
• Theta is the B allele frequency and is calculated by: (2/pi)*arctan(normalized Y/ normalized X)

You can check this math by selecting "filter rows" in the "Full Data Table" and making Theta, R, X, and Y visible for every SNP in GenomeStudio then plugging in X and Y to the formulas in 3.