Daniel Nicorici

@shenwei356 For now I am using this [extract_short_reads.py](https://github.com/ndaniel/fusioncatcher/blob/master/bin/extract_short_reads.py) for cases when the reads ids do not fit in the memory and it works well BUT it is slower than SeqTK.

@roryk I would be very carefully on filtering out found fusion genes based on blacklist filter of overlapping genes from here https://github.com/roryk/fusion-gene-blacklist . That black list contains that, for example,...

@roryk > It sounds like just flagging the fusions as possibly due to overlapping annotations/etc is a better strategy than straight up blacklisting them. I think that flagging is much...

@schelhorn > ... using starfusion we (also) often see artifactual fusion calls involving human paralogues which seem to originate from ambigious read mappings due to very high local sequence similarity...

@schelhorn >Both IGH and DUX4 are very problematic genes, to say the least. These genes are very well known and have been used as biomarkers in clinics for many years....

@schelhorn > I'm just worried that concentrating on the most difficult validated fusion gene pairs There is no such thing as **difficult** fusion gene. There is just tools/programs which are...

> Do you think it's be reasonable to compile a list of wet-lab validated fusion gene pairs, and directly test the compatibility of RNAseq reads with these "mini-chromosomes"? I think...

Ok.