2017-05-31 Bug Fix: Correct frequencies for small CNA in histograms
For a while there had been an insidious bug when determining the frequencies of particular CNA. In some image types, CNVs are binned to (usually) 1MB intervals, assigning those intervals then a value of "1" or "-1" for gains or losses, respectively. For images such as histograms showing the frequency of gains or losses ialong the genome, those 1MB intervals are then used to determine the occurrences of such events.
However, when both gain and loss CNVs occur in the same interval for the same sample, the corresponding interval value was set to a special value which should then be counted for both gains and losses. This option was broken, so that especially for some focal changes with gain/loss transition, the aberrations were detected in search, but not counted in histograms:
This is the extreme example of 204 samples with gains of CCND1, but no histogram count since the gains had a neighbouring loss, in the same 1MB interval. After the code fix, the histogram displays correctly.