Juicer import calling centromeres well outside of chromosome size.

[tolender]

• Imported contacts from Juicer .v9 and included mm10 centromeres bed. (all mouse chromosomes have centromeres from 0-3000000bp)

• Resulting object ignored supplied bed and called centromeres well outside of chromosome sizes, with zero values extending from end of actual chromosome data to the mis-calculated centromere (specifically from chr9 to chr19)

• Not sure if related but Juicer .hic has sort order chr1,2,3,4,5 etc. GENOVA object has sort order chr1,10,11,12 etc., or inter-chromosome matrices.

• Downstream analysis results in chromosome AB compartments being skewed and resulting saddle plots giving weird patterns.

• Also tried load_contacts_subset with specific regions per chromosome as a workaround but errors saying the feature has not been implemented for Juicer .hic

• Issue persists with all resolutions.

• Using centromeres=FALSE does not change the results

• Changing the centromeres to "chrom=1 start=1 end=1" in the dataframe afterwards doesn't fix it either.

Is there a temporary workaround I can use to remove the cells that are outside the bounds of the chromosomes from the dataframes?

WT.hic.40kb <- load_contacts(signal_path = 'data/WT.hic', sample_name = "WT", resolution = 40000, centromeres=mm10.centromeres, balancing = 'KR', colour = "black")

WT.hic.40kb[["CENTROMERES"]] chrom start end 1: chr1 1 1 2: chr10 4809 4809 3: chr11 9284 9284 4: chr12 13759 13759 5: chr13 18234 18234 6: chr14 22709 22709 7: chr15 27184 27184 8: chr16 31659 31659 9: chr17 36134 36134 10: chr18 40609 40609 11: chr19 45084 45084 12: chr2 49559 49559 13: chr3 54033 54033 14: chr4 57956 57956 15: chr5 61787 61787 16: chr6 65506 65506 17: chr7 69170 69170 18: chr8 72729 72729 19: chr9 75918 75918 20: chrX 79106 79106

[teunbrand]

This sounds like an off-by-1 error somewhere but I've no way of debugging this problem as I don't have mouse mm10 v9 juicer files lying around.