Exploring and Understanding how to use "cn.mops"

[sonali-bioc]

Hi Gunter,

Can you please share how you normalized the data? (Ans provided by email) - The R code can be found at - https://gist.github.com/gklambauer/8955203

Can you also explain to me about the classes used by cn.mops? eg: (CN0,CN1,...CN8)

For the output of cnvs(resCNMOPS), we get 4 metadata columns in the GRanges object. Can you elaborate about the mediate and mean columns here? How should one best interpret these?

Thanks and Regards Sonali Arora.

[gklambauer]

Normalization for tumor samples is challenging if large parts of a chromosome or even chromosome arms are duplicated. I am showing the correct normalization at https://gist.github.com/gklambauer/8955203

The copy number classes default to CN0, CN1, CN2, CN3, .., CN8. CN2 is the normal copy number for diploid samples. CN1 is a heterozygous deletion and CN0 is a homozygous deletion. CN3 thru CN8 are amplifications. For non-tumor samples the highest copy number class is 8 - higher copy numbers have not been reported. CN8 is expected to have 4 times as many reads (for times as high coverage) as CN2. For tumor samples very high copy numbers have been observed (e.g. CN64), therefore the parameters of cn.mops have to be adjusted to allow for high copy numbers. A way to set the parameters is given in https://gist.github.com/gklambauer/8955203.

[gklambauer]

The result object of cn.mops contains a slot "cnvs". In this slot the detected CNVs are stored. It gives the genomic location (chr, start, end) and then four metadata columns. These are "sampleName", which obviously gives the sample name/id, "median", "mean" and "CN". "CN" gives the estimated integer copy number of the CNV. "median" and "mean" give the median or mean individual I/NI call for this copy number segment. The individual I/NI call is something like the expected log foldchange. Log foldchanges are often used in context with copy number detection.