Titan reporting very short CNVs

[d-henness]

I'm running PhyloWGS with Titan and I'm getting an error that says I have CNVs with the same start and end locations. When I look as the segs.txt file for that sample, I see that there are indeed several CNVs with very short lengths. This seems strange to me, can anyone suggest a reason this might be happening?

[gavinha]

Hi @d-henness

My guess is that you have some noisy outliers that are being called as individual segments. You can either exclude these segments prior to PhyloWGS.

You can also try to re-run TITAN using parameters that may help to smooth over these data points. Increase the value for the argument txnExpLen.

TitanCNA/scripts/R_scripts/titanCNA.R

Lines 61 to 62 in c3061ea

	make_option(c("--txnExpLen"), type = "numeric", default = 1e15,
	help = "Expected length of segments; higher leads to longer (less sensitive) segments; float [Default: %default]"),

If you are using the snakemake workflow, you can adjust the values in config.yaml

TitanCNA/scripts/snakemake/config/config.yaml

Line 77 in c3061ea

TitanCNA_txnExpLen: 1e15