Limits...
Exploring the speed and performance of molecular replacement with AMPLE using QUARK ab initio protein models.

Keegan RM, Bibby J, Thomas J, Xu D, Zhang Y, Mayans O, Winn MD, Rigden DJ - Acta Crystallogr. D Biol. Crystallogr. (2015)

Bottom Line: AMPLE clusters and truncates ab initio protein structure predictions, producing search models for molecular replacement.Initial solutions produced by Phaser after only 5 min perform surprisingly well, improving the prospects for in situ structure solution by AMPLE during synchrotron visits.Taken together, the results show the potential for AMPLE to run more quickly and successfully solve more targets than previously suspected.

View Article: PubMed Central - HTML - PubMed

Affiliation: Research Complex at Harwell, STFC Rutherford Appleton Laboratory, Didcot OX11 0FA, England.

ABSTRACT
AMPLE clusters and truncates ab initio protein structure predictions, producing search models for molecular replacement. Here, an interesting degree of complementarity is shown between targets solved using the different ab initio modelling programs QUARK and ROSETTA. Search models derived from either program collectively solve almost all of the all-helical targets in the test set. Initial solutions produced by Phaser after only 5 min perform surprisingly well, improving the prospects for in situ structure solution by AMPLE during synchrotron visits. Taken together, the results show the potential for AMPLE to run more quickly and successfully solve more targets than previously suspected.

Show MeSH
Comparison of SHELXE CC and mean phase error for Runs 1–4 combined. Each point represents a search model and the values are either those of the successful solution or those of the highest failing CC score. Symbols distinguish SHELXE traces that do or do not exceed a mean traced chain-fragment length of 10. In all cases 15 cycles of auto-tracing were invoked in SHELXE. Each cycle included 20 iterations of density modification.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4321487&req=5

fig2: Comparison of SHELXE CC and mean phase error for Runs 1–4 combined. Each point represents a search model and the values are either those of the successful solution or those of the highest failing CC score. Symbols distinguish SHELXE traces that do or do not exceed a mean traced chain-fragment length of 10. In all cases 15 cycles of auto-tracing were invoked in SHELXE. Each cycle included 20 iterations of density modification.

Mentions: In this work, success is defined by SHELXE (Sheldrick, 2008 ▶; Usón et al., 2007 ▶; Thorn & Sheldrick, 2013 ▶) criteria: a CC value of ≥25 and a mean traced chain-fragment length of ≥10. The work of Thorn & Sheldrick (2013 ▶) suggested that a CC of ≥25 and native data to better than 2.5 Å resolution are invariably indicative of success. Since we are benchmarking against known crystal structures, mean phase errors (MPEs) for successful and failing search models can be calculated (Fig. 2 ▶). The vast majority of the cases defined as successful by the SHELXE criteria indeed have a low MPE. However, in a handful of cases, totalling only seven search models across all of Runs 1–4, solutions classified as successful have an MPE of >75°. Four of these seven false positives relate to PDB entry 2fu2 which, although reported to have a resolution limit of 2.1 Å, diffracted anisotropically to only 2.6 Å in the worst direction. This potentially explains the poor quality of the solutions despite the SHELXE statistics. PDB entry 2qyw (twice) and one search model for PDB entry 3n3f gave the other false positives, and no obvious explanation for the failure of the criteria in these cases is evident. However, three such cases in a set of 1117 (Fig. 2 ▶) is a very low failure rate of the SHELXE-based success criteria and we note that these three cases only marginally passed either the CC or the mean traced chain-length criteria.


Exploring the speed and performance of molecular replacement with AMPLE using QUARK ab initio protein models.

Keegan RM, Bibby J, Thomas J, Xu D, Zhang Y, Mayans O, Winn MD, Rigden DJ - Acta Crystallogr. D Biol. Crystallogr. (2015)

Comparison of SHELXE CC and mean phase error for Runs 1–4 combined. Each point represents a search model and the values are either those of the successful solution or those of the highest failing CC score. Symbols distinguish SHELXE traces that do or do not exceed a mean traced chain-fragment length of 10. In all cases 15 cycles of auto-tracing were invoked in SHELXE. Each cycle included 20 iterations of density modification.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4321487&req=5

fig2: Comparison of SHELXE CC and mean phase error for Runs 1–4 combined. Each point represents a search model and the values are either those of the successful solution or those of the highest failing CC score. Symbols distinguish SHELXE traces that do or do not exceed a mean traced chain-fragment length of 10. In all cases 15 cycles of auto-tracing were invoked in SHELXE. Each cycle included 20 iterations of density modification.
Mentions: In this work, success is defined by SHELXE (Sheldrick, 2008 ▶; Usón et al., 2007 ▶; Thorn & Sheldrick, 2013 ▶) criteria: a CC value of ≥25 and a mean traced chain-fragment length of ≥10. The work of Thorn & Sheldrick (2013 ▶) suggested that a CC of ≥25 and native data to better than 2.5 Å resolution are invariably indicative of success. Since we are benchmarking against known crystal structures, mean phase errors (MPEs) for successful and failing search models can be calculated (Fig. 2 ▶). The vast majority of the cases defined as successful by the SHELXE criteria indeed have a low MPE. However, in a handful of cases, totalling only seven search models across all of Runs 1–4, solutions classified as successful have an MPE of >75°. Four of these seven false positives relate to PDB entry 2fu2 which, although reported to have a resolution limit of 2.1 Å, diffracted anisotropically to only 2.6 Å in the worst direction. This potentially explains the poor quality of the solutions despite the SHELXE statistics. PDB entry 2qyw (twice) and one search model for PDB entry 3n3f gave the other false positives, and no obvious explanation for the failure of the criteria in these cases is evident. However, three such cases in a set of 1117 (Fig. 2 ▶) is a very low failure rate of the SHELXE-based success criteria and we note that these three cases only marginally passed either the CC or the mean traced chain-length criteria.

Bottom Line: AMPLE clusters and truncates ab initio protein structure predictions, producing search models for molecular replacement.Initial solutions produced by Phaser after only 5 min perform surprisingly well, improving the prospects for in situ structure solution by AMPLE during synchrotron visits.Taken together, the results show the potential for AMPLE to run more quickly and successfully solve more targets than previously suspected.

View Article: PubMed Central - HTML - PubMed

Affiliation: Research Complex at Harwell, STFC Rutherford Appleton Laboratory, Didcot OX11 0FA, England.

ABSTRACT
AMPLE clusters and truncates ab initio protein structure predictions, producing search models for molecular replacement. Here, an interesting degree of complementarity is shown between targets solved using the different ab initio modelling programs QUARK and ROSETTA. Search models derived from either program collectively solve almost all of the all-helical targets in the test set. Initial solutions produced by Phaser after only 5 min perform surprisingly well, improving the prospects for in situ structure solution by AMPLE during synchrotron visits. Taken together, the results show the potential for AMPLE to run more quickly and successfully solve more targets than previously suspected.

Show MeSH