binMissingValues: binMissingValues
In mpierrejean/jointSeg: Joint Segmentation of Multivariate (Copy Number) Signals

Description Usage Arguments Details Note Author(s) References Examples

Perform binning in order to remove missing values

1	binMissingValues(Y, verbose = FALSE)

`Y`	A numeric matrix
`verbose`	A `logical` value: should extra information be output ? Defaults to `FALSE`.

Some segmentation methods (in particular, GFLars) do not natively handle the situation when some observations have missing values in one or more dimensions. In order to avoid dropping the corresponding observations entirely, binMissingValues bins the signal values of the last complete observation before a (range of) observations with missing entries using the binMeans function.

In the specific case when the first row has NA values, the first non-missing entry is replicated in order to make smoothing possible. This choice is arbitrary but some arbitrary choice is needed in that case.

The binMissingValues function may be used for other segmentation methods suffering from the same limitation. However, we emphasize that handling missing values natively in the segmentation method would be a better solution.

Currently this function is only used by doGFLars in order to make it possible to run GFLars segmentation on SNP array data where most markers (on the order of 2/3 to 5/6) have missing values, because of uninformative or missing allelic ratio signals. The binMissingValues function may be used for other segmentation methods suffering from the same limitation. However, we emphasize that handling missing values natively in the segmentation method would be a better solution.

Morgane Pierre-Jean and Pierre Neuvial

Bleakley, K., & Vert, J. P. (2011). The group fused lasso for multiple change-point detection. arXiv preprint arXiv:1106.4199.

Vert, J. P., & Bleakley, K. (2010). Fast detection of multiple change-points shared by many signals using group LARS. Advances in Neural Information Processing Systems, 23, 2343-2351.

sim <- randomProfile(10, 1, 0.1, 3)
Y <- sim$profile
Y[c(4, 8), 2] <- NA
Y[c(7, 8), 3] <- NA

res <- binMissingValues(Y)

Y <- sim$profile
Y[1:5, 2] <- NA
Yb <- binMissingValues(Y)

Y <- sim$profile
Y[3:5, 2] <- NA
Yb <- binMissingValues(Y)