Hi-C normalization

[Natalie Sauerwald]

I had 2 questions about the VC and sqrtVC normalization methods described in the supplement of "A 3D Map of the Human Genome at Kilobase Resolution Reveals Principles of Chromatin Looping" (Cell, 2014).

1. I was unable to reproduce the exact values given in the .VCnorm and .sqrtVCnorm files - it appears that I am missing a constant, chromosome-specific factor. Could you describe the normalization factor you used to compute the values in these files?

2. In the supplement, you mention that VC norm tends to "overcorrect". Could you explain this further? What exactly is being overcorrected, and how do we know?

[Neva Durand]

Hello Natalie,

There is indeed a constant factor applied so that the sum of the matrix entries is the same.  This ensures that for all types of normalization, the total number of reads remains consistent as you change between normalizations.  In particular, in the Juicebox viewer, the color range remains consistent as you switch from flavor to flavor, and you can really see how choices of normalization affect the probability mass of the reads.

Here's the code:  https://github.com/theaidenlab/juicebox/blob/master/src/juicebox/tools/utils/original/NormalizationCalculations.java#L321

One way to see the overcorrection of vanilla coverage is to look at supplemental figure S1b.  Places where coverage was high in the original map are "overcompensated" and become depleted.  With square root, this effect is dampened, but the best remains KR.  At high resolution, we suggest VC as the alternative to KR if KR doesn't work for some reason; at low resolution, square root VC better approximates KR.

Hope that helps.

Best

Neva

--
You received this message because you are subscribed to the Google Groups "3D Genomics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3d-genomics+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/3d-genomics/dbcb8d40-ebda-4ff2-9052-244a2382cc9d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Neva Cherniavsky Durand, Ph.D.

Staff Scientist, Aiden Lab

www.aidenlab.org