Updating direct methods V. Phasing when triplet invariants are estimated via the Patterson map

Area of Science:

• Crystallography
• Structural Biology
• Computational Chemistry

Background:

• A novel probabilistic method for estimating triplet invariants using Patterson map information was previously introduced.
• Initial tests showed this method outperformed the traditional Cochran estimate, suggesting feasibility for ab initio structure solution even with lower resolution data (<2 Å).

Purpose of the Study:

• To address unresolved questions regarding the optimization of direct-methods phasing using the novel probabilistic method.
• To determine optimal parameters for Patterson peak and reflection selection, the necessity of heavy atoms, and the method's capability for ab initio phasing up to 2.2 Å resolution.
• To introduce a computer program for routine macromolecular structure solution.

Main Methods:

• Development of criteria for selecting the number of Patterson peaks for triplet estimation.
• Definition of criteria for selecting reflections in the triplet search.
• Investigation into the role of heavy atoms and the identification of enantiomorph-sensitive triplets.
• Development and testing of an automated computer program for ab initio phasing.

Main Results:

• Criteria were established for optimizing Patterson peak and reflection usage in triplet estimation.
• The necessity of heavy atoms for the successful application of the probabilistic method was confirmed.
• The method was found to identify enantiomorph-sensitive triplets but not negative cosine triplets.
• A computer program was developed that can automatically solve some test structures (proteins, nucleic acids) with data resolution up to 2.2 Å, though not without challenges.

Conclusions:

• The refined probabilistic method, with optimized parameters and the inclusion of heavy atoms, enhances ab initio phasing for macromolecular structures.
• The developed computer program demonstrates potential for routine structure solution at resolutions up to 2.2 Å, despite current limitations.
• Further development is needed to overcome limitations and fully realize the program's potential in automated structure determination.