hello, I'd like to know why do we use log2 in Copy Number Variation détection ? How does it normalize data Thank you
The word "normalize" can be slightly confusing here. You do want to normalize the probe intensity (array CGH) or read depth (NGS) ratios between test and control samples, but that's unrelated to using log scale.
Most analysis software converts copy ratios from linear to log scale so that the distribution of copy ratio values will be more Gaussian (a.k.a. normally distributed, resembling the normal distribution), and thus more amenable to statistical analysis and visualization.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Largely depends upon the tool you are using for CNV detection. Your query needs to be elaborated and the tool you are using so that we can make you understand the underlying algorithm of normalization. Just to tell you anything which is log-scale does not mean it is normalization. It means scaling. Normalization refers to either the usage of library size or normalizing factor that will be used to make the data uniform. Scaling is done mostly for the data to be represented to reduced the otherwise inflated scores. People often misconstrue scaling with normalization but two are different things. Yes when you scale for z-score calculation, then that is a different thing. So request the OP to be more clear about what tool is used and if you are asking for normalization for deriving CNVs or the scaling of the ratio scores of CNVs in log-scale?
There is additional info requirement: For somatic CNV an added normalizer might also be added other than the library size or even CBS is used.