eigenvector values flipped

[Payel Sen]

Hi, I notice that for some of our HiC samples visualized on juicebox, the eigenvector values are flipped i.e. positive values are enriched in H3K27me3 and negative values are depleted for the mark. How do you generally handle this? Is this random? How can we be sure about the designation of A and B compartments without having to go through each of the chromosomes?