writeRegions of poor SNP6 arrays failed

[thomas.g...@merckgroup.com]

Dear Henrik,

I segmented SNP6 data of a poor ddata set of ~1200 samples, first with the standard CRMA v2 analysis pipeline and Ascat ASPCF on top. This resulted in 700 failed arrays.

In order to squeeze out more information from this data set, I performed CBS within Aroma, removing all limits for possible segmentation issues, since I would like to inspect all data after the segmentation step.

>cbs <- CbsModel(ds, ref, min.width=5, maxNAFraction = 1) 

This worked for all samples.

It however lead to issues with respect to the extraction of the segmentation data:

> pathname <- writeRegions(cbs, verbose=verbose)

20190311 11:06:59|Array #1 ('ARLES_p_NCLE_DNAAffy2_S_GenomeWideSNP_6_A01_256002') of 1213...

20190311 11:06:59| Extracting regions from all fits...

20190311 11:06:59|  Obtaining CN model fits (or fit if missing)...

Loading required package: DNAcopy

Error in sort(abs(diff(genomdat)))[1:n.keep] : 

  only 0's may be mixed with negative subscripts

Is there a workaround for this, e.g. is it possible to inspect the segmentation data with different thresholes for %of segmentation failures?

rgds

Thomas