Dataset Information

Validation of experimental charge-density refinement strategies: when do we overfit?

ABSTRACT: A cross-validation method is supplied to judge between various strategies in multipole refinement procedures. Its application enables straightforward detection of whether the refinement of additional parameters leads to an improvement in the model or an overfitting of the given data. For all tested data sets it was possible to prove that the multipole parameters of atoms in comparable chemical environments should be constrained to be identical. In an automated approach, this method additionally delivers parameter distributions of k different refinements. These distributions can be used for further error diagnostics, e.g. to detect erroneously defined parameters or incorrectly determined reflections. Visualization tools show the variation in the parameters. These different refinements also provide rough estimates for the standard deviation of topological parameters.

SUBMITTER: Krause L

PROVIDER: S-EPMC5571805 | biostudies-literature | 2017 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Validation of experimental charge-density refinement strategies: when do we overfit?

Krause Lennard L Niepötter Benedikt B Schürmann Christian J CJ Stalke Dietmar D Herbst-Irmer Regine R

IUCrJ 20170524 Pt 4

A cross-validation method is supplied to judge between various strategies in multipole refinement procedures. Its application enables straightforward detection of whether the refinement of additional parameters leads to an improvement in the model or an overfitting of the given data. For all tested data sets it was possible to prove that the multipole parameters of atoms in comparable chemical environments should be constrained to be identical. In an automated approach, this method additionally ...[more]

PMID: 28875029

Similar Datasets

Project description:The pharmaceutical agent theophylline (THEO) is primarily used as a bronchodilator and is commercially available in both tablet and liquid dosage forms. THEO is highly hygroscopic, reducing its stability, overall shelf-life, and therefore usage as a drug. THEO and dicarboxylic acid cocrystals were designed by Trask et al. in an attempt to decrease the hygroscopic behaviour of THEO; cocrystallisation of THEO with malonic acid (MA) did not improve the hygroscopic stability of THEO in simulated atmospheric humidity testing. The current study employed high-resolution X-ray crystallography, and Density Functional Theory (DFT) calculations to examine the electron density distribution (EDD) changes between the cocrystal and its individual components. The EED changes identified the reasons why the THEO:MA cocrystal did not alter the hygroscopic profile of THEO. The cocrystal was equally porous, with atomic packing factors (APF) similar to those of THEO 0.73 vs. 0.71, respectively. The THEO:MA (1) cocrystal structure is held together by an array of interactions; a heterogeneous synthon between the imidazole and a carboxylic fragment stabilising the asymmetric unit, a pyrimidine-imidazole homosynthon, and an aromatic cycle stack between two THEO moieties have been identified, providing 9.7-12.9 kJ mol-1 of stability. These factors did not change the overall relative stability of the cocrystal relative to its individual THEO and MA components, as shown by cocrystal (1) and THEO being equally stable, with calculated lattice energies within 2.5 kJ mol-1 of one other. The hydrogen bond analysis and fragmented atomic charge analysis highlighted that the formation of (1) combined both the EDD of THEO and MA with no net chemical change, suggesting that the reverse reaction - (1) back to THEO and MA - is of equal potential, ultimately producing THEO hydrate formation, in agreement with the work of Trask et al. These results highlight that a review of the EDD change associated with a chemical reaction can aid in understanding cocrystal design. In addition, they indicate that cocrystal design requires further investigation before becoming a reliable process, with particular emphasis on identifying the appropriate balance of synthon engineering, weak interactions, and packing dynamics.

Dataset Information

Validation of experimental charge-density refinement strategies: when do we overfit?

Publications

Validation of experimental charge-density refinement strategies: when do we overfit?

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets