Hello,
In the
CODEX2 tutorial, there are recommended post-segmentation filters listed at the end of the tutorial:
"Post-segmentation pruning and filtering are recommended based on CNV
length (filter1), length per exon (filter2), likelihood ratio (filter3),
and number of exons (filter4)."
```
filter1 <- finalcall.CBS$length_kb<=200
filter2 <- finalcall.CBS$length_kb/(finalcall.CBS$ed_exon-finalcall.CBS$st_exon+1)<50
finalcall.CBS.filter <- finalcall.CBS[filter1 & filter2, ]
filter3 <- finalcall.CBS.filter$lratio>40
filter4 <- (finalcall.CBS.filter$ed_exon-finalcall.CBS.filter$st_exon)>1
finalcall.CBS.filter=finalcall.CBS.filter[filter3|filter4,]
```
I am curious, how were these filters chosen/how were the thresholds determined? Is there any biological significance behind them?
I have been using CODEX2 to call CNVs and am unsure of whether or not I need to apply these filters and/or adjust the threshold values.
Thanks,
Layne