Transforming Beta values to M values

Question

[ChIP]

Hi,

I have a set of beta values and I need to transform these into M-values, the formula is quite straight forward:

M=log2(Beta/(1-Beta))

Now, the problem is some of my Beta values have NA, how to deal with these NA tags while performing this transformation?

An example set that I have is

Composite_Element_REF Beta_value Gene_Symbol Chromosome Genomic_Coordinate
cg00000029 0.8265891628 RBL2 16 53468112
cg00000108 NA C3orf35 3 37459206
cg00000109 NA FNDC3B 3 171916037
cg00000165 0.1479752852 1 91194674

For first and last one it is straight forward calculation of M-value but how can I calculate it for 2 and 3 value.

Any suggestions, if you have dealt with this situation, FYI this data has been taken from TCGA level3.

Thank you

Answer 1

[Devon Ryan]

I would suggest simply omitting those points. They're likely NA because of a problem during data processing (e.g., no signal). The only other time this can happen is if the beta values were calculated from the M values (fairly likely) and that hit +/- infinity. How that's handled ends up being dependent on exactly how the data was processed, so you can't reliably just replace NA with 0 or 1 without knowing that. You might try to contact the work group that produced that particular TCGA dataset and just ask them about the NAs.

[ChIP]

Hi Devon,

I agree with you, but the problem with omitting these points is that these are not consistent over a set. Each set has different data points that have NA.

But thanks for your answer.