I am using bedtools fisher as follows:
bedtools fisher -a $fA -b $fB -g $fG > $fOut
Where fA consists of wide peaks and fB narrower peaks. The results are:
Number of query intervals: 741 Number of db intervals: 18079 Number of overlaps: 1481 Number of possible intervals (estimated): 146318 phyper(1481 - 1, 741, 146318 - 741, 18079, lower.tail=F) Contingency Table Of Counts _________________________________________ | in -b | not in -b | in -a | 1481 | 0 | not in -a | 16598 | 128239 | _________________________________________ p-values for fisher's exact test left right two-tail ratio 1 0 0 inf
in -a in -b (1481) is fine because several of fB peaks may fall within one fA peak. However, there are about 150 fA peaks which do not have any overlap with fB (as found using
bedtools intersect (counting unique ID only) and also verified in a browser).
So what is the meaning of cell
in -a not in -b showing zero ?