Calibrating using pureCN causes multiple segments to go -10

[shashwatsahay]

Calibrating segments with pureCN output causes output to have many segement copy ratio to go to -10

uncalibrated

calibrated

Also For a particular sample I have oversegmentation any ideas how to resolve this

[shashwatsahay]

Have solved the -10 log values these are due to me not dropping segments with low coverage.

Any recommendation fo noisy sample

[etal]

The usual ways to reduce noise are:

• Refine the reference profile -- ideally use 10-30 good-quality normal samples, a reference genome FASTA, and an exclusion list of known problematic (e.g. unmappable) regions. After using the reference on some known positive control (test) samples, see if any additional regions are simply not performing well, and delete these rows from the reference .cnn file, too.
• Implement some QC checks on the test/tumor samples -- some sequenced samples may be too degraded or shallow (though I don't think that's the case here)
• For segmentation, consider --drop-low-coverage if you have tumor samples. Beyond that, perhaps try tuning outlier smoothing, or a different segmentation method -- active research here, not usually the most promising route if you're just trying to get results.
• Apply more stringent post-filters with the segmetrics and call commands -- usually the easiest way to rescue a dataset that has only minor issues, if the reference profile is already in good shape.