search

gersteinlab / HiC-spector

Spectral and reproducibility analysis of Hi-C contact maps
10 stars 7 forks source link

Eigenvector normalization #9

Closed srcoulombe closed 3 years ago

srcoulombe commented 3 years ago

The paper qualifies the eigenvectors used in computing the distance metric as being 'normalized', but I don't see where nor how that normalization is done in the Python code. My guess is the norm is implied to be the L2 norm so that the distance metric compares pairs of unit-length vectors, but that doesn't seem to be the case in the Python implementation, but that doesn't seem to be the case in the Python implementation.

quantum-man commented 3 years ago

It's what you have guessed. I think by default eigsh returns unit eigenvectors.

On Tue, Mar 16, 2021 at 9:55 AM srcoulombe @.***> wrote:

The paper qualifies the eigenvectors used in computing the distance metric as being 'normalized', but I don't see where nor how that normalization is done in the Python code. My guess is the norm is implied to be the L2 norm so that the distance metric compares pairs of unit-length vectors, but that doesn't seem to be the case in the Python implementation, but that doesn't seem to be the case in the Python implementation.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/gersteinlab/HiC-spector/issues/9, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABZKLPABUI2JWLSA7PY4C5TTD5WMTANCNFSM4ZIWXUAQ .

srcoulombe commented 3 years ago

Yes OK thanks for clarifying