Skip to content

Code for DMVGAE. We will release codes after ICASSP 2023.

Notifications You must be signed in to change notification settings

bozhenhhu/DMVGAE

Repository files navigation

DMTPRL

Motivation:

The geometric consistency of AlphaFold2 is maintained on the high-dimensional edge representations.

Geoformer could optimize high-dimensional representations to capture complex interaction patterns among amino acids while still maintaining geometric consistency in Euclidean space. In each Geoformer layer, these embeddings are updated iteratively to refine the geometric inconsistency: each node embedding was updated with related pairwise embeddings, and each pairwise embedding was updated by triangular consistency of pairwise embeddings.

However, most protein methods learned protein representations are usually not constrained, leading to performance degradation due to data scarcity, task adaptation, etc. Can we design a loss to satisfy the demand for geometric consistency?

Main codes

for Deep Manifold Graph Auto-Encoder For Attributed Graph Embedding and Deep Manifold Transformation for Protein Representation Learning

Codes are based on DMT and KeAP.

KeAP: paper and codes

DMT: paper and codes

About

Code for DMVGAE. We will release codes after ICASSP 2023.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published