Pseudoracemic amino acid complexes: blind predictions for flexible two-component crystals

Ab initio prediction of the crystal packing in complexes between two flexible molecules is a particularly challenging computational chemistry problem. In this work we present results of single crystal structure determinations as well as theoretical predictions for three 1 : 1 complexes between hydrophobic L-and D-amino acids (pseudoracemates), known from previous crystallographic work to form structures with one of two alternative hydrogen bonding arrangements. These are accurately reproduced in the theoretical predictions together with a series of patterns that have never been observed experimentally. In this bewildering forest of potential polymorphs, hydrogen bonding arrangements and molecular conformations, the theoretical predictions succeeded, for all three complexes, in finding the correct hydrogen bonding pattern. For two of the complexes, the calculations also reproduce the exact space group and side chain orientations in the best ranked predicted structure. This includes one complex for which the observed crystal packing clearly contradicted previous experience based on experimental data for a substantial number of related amino acid complexes. The results highlight the significant recent advances that have been made in computational methods for crystal structure prediction.