search

dozmorovlab / HiCcompare

Joint normalization of two Hi-C matrices, visualization and detection of differential chromatin interactions. See multiHiCcompare for the analysis of multiple Hi-C matrices
https://dozmorovlab.github.io/HiCcompare/
Other
19 stars 3 forks source link

the data is mismatch? #5

Closed RuiqinZheng closed 6 years ago

RuiqinZheng commented 6 years ago

head(HMEC.chr22) region1 region2 IF 1 16000000 16000000 5 2 16000000 16500000 2 3 16500000 16500000 297 4 16000000 17000000 5 5 16500000 17000000 92 6 17000000 17000000 5690

head(NHEK.chr22) region1 region2 IF 1 16000000 16000000 6 2 16000000 16500000 10 3 16500000 16500000 554 4 16000000 17000000 17 5 16500000 17000000 320 6 17000000 17000000 10880 head(chr22.table) chr1 start1 end1 chr2 start2 end2 IF1 IF2 D 1: chr22 16000000 16500000 chr22 16000000 16500000 5 5.203323 0 2: chr22 16000000 16500000 chr22 16500000 17000000 2 8.672206 1 3: chr22 16500000 17000000 chr22 16500000 17000000 297 480.440193 0 4: chr22 16000000 16500000 chr22 17000000 17500000 5 14.742750 2 5: chr22 16500000 17000000 chr22 17000000 17500000 92 277.510581 1 6: chr22 17000000 17500000 chr22 17000000 17500000 5690 9435.359752 0 M 1: 0.05750528 2: 2.11639897 3: 0.69389392 4: 1.56000562 5: 1.59283701 6: 0.72964887

you see ,1) the original IF2 should be 6,but why the IF2 is 5.203323 in hic.table? 2) in my own data the first row of first data is: 1 1 624.0 the first row of second data is: 1 1 821.0 but why in the hic.table like this? chr1 start1 end1 chr2 start2 end2 IF1 IF2 D M
1: chr1 1 2 chr1 1 2 624 572.9156 0 -0.123223336 Does the software have do something when I use create.hic.table? another question is can I use this package to caculate inter-chromosme different interacion? Best Wishes~ Thanks a lot!

jstansfield0 commented 6 years ago

The data is not a mismatch. Scaling is performed by default in create.hic.table. You can turn it off if you set scale = FALSE.

HiCcompare is not designed for looking at inter-chromosomal differences. Most inter-chromosomal contact maps are very sparse and noisy which make them hard to analyze.

RuiqinZheng commented 6 years ago

oh,I see it also can caculate the difference bwteen different-chromsome right? but the results maybe not crediable?

jstansfield0 commented 6 years ago

I think you will get errors if you try to use inter-chromosomal data due to the fact that distance cannot be calculated between different chromosomes.

RuiqinZheng commented 6 years ago

yeah you are right.but if I just consider the all chromosome interactions as a big chromsome interaction? and if I set the scale=FALSE,does it will influence the final result?

jstansfield0 commented 6 years ago

I would not recommend combining all the data into a single matrix. It most likely will not work as this is not the intended use of HiCcompare.

Setting scale = FALSE just changes if there is an initial scaling based on total number of reads. The loess normalization should be able to account for total number of reads as well and scaling is not always necessary.

RuiqinZheng commented 6 years ago

OK! Thanks a lot !