Closed yeroslaviz closed 7 years ago
I have never looked closely at the bigwigs. It is not intentional; I'll try to figure out why this happens, but I do not have the time within the next two weeks. However, the results are still based on data that is overlapping, without any gaps.
Can you show me how you read the data (which libraries and example code) and I'll get around to debugging this sometime in the future.
Thanks for reporting 👍
reading the file with epic
epic --control sizKO.input.0h.1.reads.bedpe --treatment sizKO.ip.0h.1.reads.bedpe -cpu 10 --genome sacCer3 -egf 0.961651807729 -cs Sc.R64.1.genome -b bedFiles/sizKO.InputIP.0h.1.bed -sm matrixFiles/sizKO.InputIP.0h.1.matrix -bw bigwigFiles/sizKO.InputIP.0h.1 -cbw ChipBigwigFiles/sizKO.InputIP.0h.1 --outfile OutputFiles/sizKO.InputIP.0h.1.Output --log logFiles/sizKO.InputIP.0h.1.log -fdr 0.1 -w 150
reading it into R
library(rtracklayer)
ip.file <- import.bw(con = "../EpicResults/sizKO.ip.0h.1.reads_log2fc.bw")
... However, the results are still based on data that is overlapping, without any gaps.
I am not sure what this means regarding the results i have. If they are overlapping, the positions are completely wrong. if the first bin ends at the 149 and the second begins at 200, I have a gap of 50 bases, if the second bin begins at 150 (or 151 if the structure is consistent), the given positions are shifted 50 bases for each bin. This would cause a huge shift at the end of each chromosome. It still looks like it is based on a 200 bases windows, but calculate the coverage for only 150 bases within this 200 bases each time.
Great catch, I did not see that part! For the -w 150 the results are probably very wrong!
I'll try to find some time to start looking at this tomorrow. It is a serious error.
Endre
I can reproduce the 150-error for paired-end data, but not single-end beds. This error also shows up in the matrix creation (--sm
).
I thanked you in the CHANGELOG. Should be fixed in 0.2.5. The problem was simply a hardcoded bin-size. Thanks again for reporting 👍
There is something weird about the results here. When I am reading the resulted bigwig file (i2bw file) into my R session, I am getting non-overlapping bins with one position skipped each time.
It looks line each time one base is skipped from the counting (200,400,600,etc.) It is even worse when the -w parameter is set to 150.
Here each time 51 bases are skipped.
Is this deliberate? I hope it is not to confusing, that I show the R output here, as this is how I can visualize my bigwig results.