Description
Hi-C is a method for capturing contacts between all pairs of DNA fragments in one
experiment. In the process of Hi-C experiment, pairs of fragments appearing as ligation are
recorded as interactions. However, not all of these ligations are due to real meaningful interactions
between DNA fragments. Some may have been generated as a result of random collisions due to
Brownian motion of fragments in the solution. These pairs of fragments may be close to each other
in DNA, or may have come close to each other during dynamic movements of DNA. The goal of
background correction models is to calculate the significance of the recorded interactions for
distinguishing between real and random interactions. Max-Hi-C is a statistical background model
for both general Hi-C and capture Hi-C libraries. The significance of each interaction is calculated
as a function of bias factors of its two ends and their genomic distance. The training is done by
maximizing the likelihood of the observed interactions.
