Hydrogen bond strengths for large molecules by local magnetically induced currents

The objective of this research project is to develop novel computational methods for estimating hydrogen-bond strengths in large molecular systems using local magnetically induced currents. Computationally, magnetically induced currents are calculated in two steps using the gauge including magnetically induced current (GIMIC) method. First a nuclear magnetic resonance (NMR) shielding calculation has to be done where necessary input information is generated. This can be done using any quantum chemistry software that is interfaced to GIMIC. We use LSDalton for this purpose. The NMR step is very important because the calculated NMR chemical shifts can directly be compared to experimental measurements. Then the magnetically induced currents are calculated in a second step using GIMIC. Both programs (GIMIC and LSDalton) are open source projects. This means that new developments are freely shared with the scientific community. So far, we have a working version of the code that is able to calculate specific NMR shieldings only for those nuclei or areas of a molecule we are really interested in. That means we can avoid a huge amount of unnecessary computations of shieldings. This achievement is part of the latest LSDalton release and made free available to the scientific community. Thus, the first step for our goal to develop local currents has been achieved but we face different technical problems related to the scaling behavior of our code. Therefore, we extended our implementation and included a method called density-fitting to speed up computationally expensive Coulomb integral calculations. The largest molecule we tested consisted of 1400 atoms, which is very promising and shows that our efforts go in the right direction. The usage of density-fitting improved the scaling behavior significantly but it is still not optimal. Therefore, we investigated other techniques such as the auxiliary-density matrix method (ADMM) to speed up the calculation of exchange integrals to improve the present situation. This works now in LSDalton for ground state energies, excitation energies and electric response properties such as polarizabilities and hyperpolarisabilities, which is a big step forward. However, the inclusion of the ADMM method for the calculation of polarizabilities and hyperpolarizabilities turns out to be challenging since the performance gain comes at a cost of accuracy indicating further challenges upon extension to magnetic properties, which is needed to calculate NMR shieldings. On the other hand, we need to improve the GIMIC software such that it is able to handle the new and data intensive NMR information provided by LSDalton to get access to the current density. Therefore, GIMIC has to be optimized such that it operates efficiently and highly parallel on a supercomputing architecture providing another challenge. In this context we were able to acquire an advanced user support project (AUS) with Uninett/Sigma2 the Norwegian Supercomputing Centre. Sigma2 is giving us access to computer science know how, needed to perform our tasks. As a result, the GIMIC software is currently completely restructured. The code is adapted to work OpenMP, which means optimized performance on one node with 20 CPUs, which is an important improvement compared to a serial code running on one node and one CPU only. An automated testing procedure via Travis CI has been introduced as quality measure, assuring that new code contributions do not affect working ones. A very encouraging outcome of this project is that we got an invited advanced review article on magnetically induced current densities accepted in a high rank international journal demonstrating a growing interest in this research field. An invited book chapter on the same topic, discussing porphyrin compounds and current densities got published in 2018 as part of a book series of the Royal Chemical Society, indicating the growing relevance of this research field to organic chemists. It turned out that analysing magnetically induced currents provides additional information that is relevant for the understanding and interpretation of unusual NMR spectra making the method interesting for experimentalists. As a consequence, we get more and more requests from experimental working groups to complement their spectroscopic measurements by a theoretical current density study. One of these cooperative works got published as a communication article in a highly ranked international journal, increasing the visibility of our group and this research field. As a consequence, our group is facing more collaboration requests than we are able to accept. Therefore, we need to provide more user support and training to increase the applicability and user community of our software.

-

The objective of this research project is to develop novel computational methods for estimating hydrogen-bond strengths in large molecular systems using local magnetically induced currents. Hydrogen-bonds play an important role in biomolecular systems. T hey account for DNA base pairing and are involved in many enzymatic processes. Knowledge about hydrogen bond energetics is important for understanding (i) protein structures, (ii) enzymatic activities. Proton transfer (pT) plays a crucial role regarding e nzyme catalysis and transport reactions where the proton has to be transported along a significant distance. For example, hydrogen bonded water molecules form the basis for rapid proton transport in biological systems, such as cytrochrome c oxidase a proc ess which is not yet fully understood. The realization of the proposed research project can be expected to establish a high quality new research field. The described unconventional approach for estimating hydrogen bond strengths is new and demonstrates al ternative ways of thinking. Knowledge of hydrogen bond strength of large biomolecules will lead to a better understanding of weak interactions in large molecular systems and thus open new opportunities for understanding enzymatic and biochemical process es. This might lead to new and more efficient drugs, which implies high relevance for society national and international.