Use this URL to cite or link to this record in EThOS:
Title: Enhanced structure determination from powder diffraction data via algorithm optimisation and the use of conformational information
Author: Kabova, Elena A.
ISNI:       0000 0004 6347 0106
Awarding Body: University of Reading
Current Institution: University of Reading
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
The performance of DASH has been evaluated against powder X-ray diffraction data collected from 101 molecular crystal structures, representing the most comprehensive testing of a "structure determination from powder diffraction data" (SDPD) program carried out to date. These 101 structures cover a broad range of molecular complexities, from very simple (6 degrees of freedom) to very challenging (49 degrees of freedom). 95 of the crystal structures could be solved with the current version of DASH, going some way to explaining why the parameterisation of its simulated annealing (SA) algorithm has not been altered since the launch of the program in 1999. This thesis explores optimisation of key DASH SA parameters using the program irace. The irace runs, comprising 255,000 individual DASH runs and requiring approximately 1300 CPU days of compute time, produced six sets of SA parameters which differed greatly from the DASH default parameters and which markedly improved the performance of DASH. Further evaluation of these six sets against all 101 compounds (a further 2874 of days of CPU time), allowed selection of one best-performing set, which delivered an order of magnitude improvement in the success rate with which crystal structures were solved. The adoption of these parameter values as the defaults in future releases of DASH is strongly recommended and is expected to broaden the range of molecular complexities to which the program can be applied. Three distinct approaches to further improving DASH performance, based on introducing prior conformational knowledge derived from the Cambridge Structural Database (CSD), have also been assesed. The findings show that inclusion of conformational knowledge brings significant additional gains in SDPD performance, and that existing implementations of these approaches in the DASH / CSD System are close to being ready for routine use.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available