Adding hydrogen atoms to molecular models via fragment superimposition

Abstract

Background Most experimentally determined structures of biomolecules lack annotated hydrogen positions due to their low electron density. However, thorough structure analysis and simulations require knowledge about the positions of hydrogen atoms. Existing methods for their prediction are either limited to a certain range of molecules or only work effectively on small compounds. Results We present a novel algorithm that compiles fragments of molecules with known hydrogen atom positions into a library. Using this library the method is able to predict hydrogen positions for molecules with similar moieties. We show that the method is able to accurately assign hydrogen atoms to most organic compounds including biomacromolecules, if a sufficiently large library is used. Conclusions We bundled the algorithm into the open-source Python package and command line program . Since usually no additional parametrization is necessary for the problem at hand, the software works out-of-box for a wide range of molecular systems usually within a few seconds of computation time. Hence, we believe that could be a valuable tool for structural biologists and biophysicists alike.

Biotite: new tools for a versatile Python bioinformatics library

“…Hydride ( https://hydride.biotite-python.org/ ) [ 65 ] is a package to predict hydrogen positions for those structure models, in which hydrogen atoms could not resolved experimentally. This enables hydrogen bond measurement and accurate base pair identification on a wider range of structures.…”