# of discordant reads

[hj6-sanger]

Dear Staff.

I appreciate you for developing and maintaining this wonderful tool.

I have a quick question.

The number of discordant reads in *_edges_cnseg.txt (I know the number is double counted) is different from the number in *graph.txt.

Some discordant reads in *_edges_cnseg.txt seem to disappear in *graph.txt.

Could you tell me why? (I guess discordant reads that are not supported by CNVs are removed?)

Thanks again for developing this wonderful tool.

BW

[jluebeck]

Hi BW,

This is a very good question, and @virajbdeshpande may better be able to answer, but I believe it occurs when the edge connects to something outside the focally amplified areas detected by AA. These may be biologically interesting (e.g. possible HSR integration points), but they may also be possible false-positive SVs that are essentially filtered from the graph file. The other, and in my opinion, less likely alternative is that the region they join to was mistakenly not detected as amplified by AA and was thus not included in the focal amplification.

Thanks, Jens