Apr 8, 2020

End to end differentiable protein structure refinement

BioRxiv : the Preprint Server for Biology
Pu Tian, X. Cao

Abstract

Refinement is an essential step for protein structure prediction. Conventional force fields (both physics and knowledge-based ones) and sampling (molecular dynamics simulations and Monte Carlo algorithms) based methods are computationally intensive on the one hand, and are labor intensive for updating parameters on the other hand. A number of neural network based methods have been developed for prediction of global protein structures. However, no differentiable refinement algorithm is available up to date. Based on neural network implementation of local maximum likelihood approximation of generalized solvation free energy theory, we develop a fully differentiable refinement algorithm with clear physical interpretation. Instead of explicit functions utilized by conventional force field approach, molecular interactions are described by neural network parameters, updating of which may be readily realized by training. Substitution of configuration sampling by differentiation increases optimization efficiency by many orders of magnitude. Additionally, both global and local conformation restraints are added to further improve the refinement algorithm. More importantly, due to modular separation of coordinates transformation (updating...Continue Reading

  • References
  • Citations

References

  • We're still populating references for this paper, please check back later.
  • References
  • Citations

Citations

  • This paper may not have been cited yet.

Mentioned in this Paper

Computer Software
Extracellular Matrix
Bio-Informatics
Codon Genus
Codon (Nucleotide Sequence)
Nucleotides
License
Simulation
Active ingredient
Biological Evolution

Related Feeds

BioRxiv & MedRxiv Preprints

BioRxiv and MedRxiv are the preprint servers for biology and health sciences respectively, operated by Cold Spring Harbor Laboratory. Here are the latest preprint articles (which are not peer-reviewed) from BioRxiv and MedRxiv.

Bioinformatics in Biomedicine (Preprints)

Bioinformatics in biomedicine incorporates computer science, biology, chemistry, medicine, mathematics and statistics. Discover the latest preprints on bioinformatics in biomedicine here.