I installed LOLA 1.4.0 in R 3.3.2 and Bioconductor 3.4 by biocLite("LOLA"). It seems that readBed() and loadRegionDB() would create GRanges without adding 1 to the left sites in the bed files, leading to not so right 1-based GRanges which would latter give out not so right results.
I find that runLOLA() invokes countOverlaps(). However, countOverlaps() is based on 1-based closed interval. So readBed() would in fact add one more base on the left side of all the regions. Even it would have small impact on the result, it maybe better to be fixed. :)
You are right.
BED format is 0-based, but countOverlaps is 1-based, so import.bed is doing it correctly and readBed is not. As you say, for LOLA probably not a big deal, but I will fix this. Thanks!
Hi Nathan,
I installed LOLA 1.4.0 in R 3.3.2 and Bioconductor 3.4 by
biocLite("LOLA")
. It seems thatreadBed()
andloadRegionDB()
would create GRanges without adding 1 to the left sites in the bed files, leading to not so right 1-based GRanges which would latter give out not so right results.I find that
runLOLA()
invokescountOverlaps()
. However,countOverlaps()
is based on 1-based closed interval. SoreadBed()
would in fact add one more base on the left side of all the regions. Even it would have small impact on the result, it maybe better to be fixed. :)