The seventh blind test of crystal structure prediction: structure generation methods

Abstract

A seventh blind test of crystal structure prediction was organized by the Cambridge Crystallographic Data Centre featuring seven target systems of varying complexity: a silicon and iodine-containing molecule, a copper coordination complex, a near-rigid molecule, a cocrystal, a polymorphic small agrochemical, a highly flexible polymorphic drug candidate, and a polymorphic morpholine salt. In this first of two parts focusing on structure generation methods, many crystal structure prediction (CSP) methods performed well for the small but flexible agrochemical compound, successfully reproducing the experimentally observed crystal structures, while few groups were successful for the systems of higher complexity. A powder X-ray diffraction (PXRD) assisted exercise demonstrated the use of CSP in successfully determining a crystal structure from a low-quality PXRD pattern. The use of CSP in the prediction of likely cocrystal stoichiometry was also explored, demonstrating multiple possible approaches. Crystallographic disorder emerged as an important theme throughout the test as both a challenge for analysis and a major achievement where two groups blindly predicted the existence of disorder for the first time. Additionally, large-scale comparisons of the sets of predicted crystal structures also showed that some methods yield sets that largely contain the same crystal structures.

Keywords: Cambridge Structural Database; blind test; crystal structure prediction; lattice energy; polymorphism.

Figure 1

Crystal packing of XXVII Form A at 90 K excluding the bromine atoms from an impurity, and highlighting the two TIPS groups: A (orange) and B (blue). Hydrogen atoms were omitted for clarity.

Figure 2

Stacked molecules of XXVIII, viewed along the crystallographic c axis, showing the weak intermolecular H⋯Cl hydrogen bonds detected at a distance of the sum of van der Waals radii plus 0.1 Å.

Figure 3

Molecules in the asymmetric unit of XXX Form A, highlighting the disorder observed in the alkyl chain.

Figure 4

Crystal packing motif of XXXIII Form A (top) and Form B (bottom). Hydrogen atoms were omitted for clarity.

Figure 5

Molecular dynamics (MD)-based analysis of molecule XXVII conformational ensemble at finite temperature. (a, top) Free energy surfaces corresponding to TIPS A (orange) and B (blue) obtained by biasing the angles ϕ₁ and ϕ₂ shown in the middle. These show different behaviour in both basin shapes and locations. (a, bottom) Equilibrium probability distributions derived from the free energy surfaces with bounding boxes used to calculate the equilibrium probability of each conformational state. These were further divided into three regions over ϕ₁ to account for the configuration of the three isopropyl groups. This results in equilibrium probabilities reported in panels (b) and (c), where the molecular structure of the main conformers is shown associated with the colour corresponding to the appropriate bounding box.

Figure 6

An overlay of the two polytypes of XXIX Form A; structures ranked first (light green, P2₁/c) and second (violet, Pc) submitted by Group 20. Note how every sixth layer is oriented differently in the two polytypes. Hydrogen atoms are omitted for clarity.

Figure 7

On top is the artificial target PXRD pattern given to participants, shown here without background profile. Second from the top is a pattern simulated from the experimental single-crystal X-ray diffraction (SCXRD) structure of Form A (P2₁/c). The blue pattern, third from top, corresponds to the closest matching predicted structure after its lattice parameters have been adjusted with VC-PWDF. The red pattern, fourth from top, was simulated from the same crystal structure as found by CSP by Group 20. The bottom two patterns correspond to a polytype structure in space group Pc found by Group 10, as found by CSP (green), and after optimizing the PXRD similarity (orange). Note the subtle differences in Bragg peak positions and extinctions between the Pc and P2₁/c structures. Inserted below is a PXRD intensity difference plot of the lattice-adjusted CSP structures relative to the SCXRD structure of Form A. The y axis has been scaled by a factor of 5 to aid the eye.

Figure 8

Convex hull (grey line) of the free energy of formation, approximated as the PBE-D3 lattice energy, Δ_fE, of cannabinol tetramethylpyrazine cocrystals as a function of their composition. The data was provided by Group 22. Each data point corresponds to a distinct predicted cocrystal structure. Note that structures of three different stoichiometries lie on the convex hull.

Figure 9

COMPACK overlay of XXXII Form B at 90 K (coloured by element) with the redetermination from PXRD of XXXII Form B (ambient temperature) by Group 20 (coloured green). Hydrogen atoms were omitted for clarity.

Figure 10

Crystal structure set similarity heat maps for molecules XXVII and XXIX showing the percentage of structures from the group on the horizontal axis that match a structure from the group on the vertical axis. Top plots show the 100 versus 100 comparisons, while those at the bottom the 100 versus all ones. Some groups have to a large extent predicted the same crystal structures. The comparisons are not symmetric because multiple structures in one set can match a single structure from another one; this is possibly due to stricter clustering criteria.