How to filter #121
Description
Dear Andy,
I have another question about Brass.
I used it for 15 short read WGS cases (tumor depth 60x and non-tumor depth 30x) and would like to filter the data to filter out as many false positive entries as possible.
The first port of call is of course the read count (for translocations between 4 and 30). If I set a VAF of >= 10 here (i.e. at least 6), I filter out many translocations that would make genetic sense for this tumor, but keep some entries that are definitely false positives, because you can see manually that this entry has the same breakpoint in a repetitive region in all 15 cases.
Therefore, I would like to switch to other filter criteria, but I don't really understand the Wiki annotation.
How exactly should I understand the occL and occH entries? Does that mean that the more events I have in occL, the more certain the entry is? Or are variants most credible when occL and occH have the same value? Because in the variant above, which is very likely to be a false positive, occL is at 8, which is the highest value.
I would also be happy if you could give me a filter recommendation (also for other headers such as copynumber_flag).
I would like to validate the things very quickly and then publish them.
Thank you in advance for your help (and your help with my other problem) and this great tool!
Best wishes
Lilian