1. Complete automation in CCP4 What do we need and how to achieve it?

2. Starting point
Given data and experimental setup (protein name or sequence, crystallisation condition(s), probable ligands) to solve the structure, complete refinement with complete model building, fit ligand(s), validate (against experiment as well as prior knowledge), do early analysis or calculations to help early analysis.
All calculations and decisions should be made automatically
I estimate success rate using current technology roughly around 80% (may be more).

3. Overall view Interpretation DATA Optimisation feedback data
MR/heavy atom/ab in
Interpretation
DATA
Optimisation
feedback data
Critical eval
Validation
Completion
Early analysis

4. Aim: Automation of X-ray Analysis (Equivalent to Microscope)
Data
Data base of
chemical,
structural
knowledge
Molecular
replacement
Experimental
phasing
Refine/Build/Rebuild
Refine/Complete !
Final structure

5. Knowledgebase: Discuss
Structural knowledge
domains, multimers, rotamers, links between similar structures, fragments and their properties.
Crystals - cells, intensities, solvent contents, resolutions …
Chemical knowledge
Monomers and their valence bond chemistry
Possible modifications
Smaller structural units
Protocols and structure solution techniques
How to organise them? How to use them?

6. Components
DATA – Experimental data, sequence/protein name, crystallisation condition(s), ligand(s)
Interpretation – Using DB and accumulated knowledge build the model
Optimisation - Refine the model against data and prior knowledge
Criticism – Is model consistent with the data and prior knowledge. Some of the prior knowledge may have not been used in optimisation
Feedback data – Electron density map, information about misfitted parts, new prior information taken from the DB
Completion – Tidying up waters, multiple conformation, hydrogen bonding network. Are there heavier atoms. Etc
Validation – Is model consistent with data and prior knowledge. More sophisticated analysis
Early analysis – Is there similar structures, family of proteins, similarity of ligand binding etc.

7. Information flow
What is needed by the next stage? In the absence of previous
step how to generate necessary information? CCP4i works
partially deals with this.

8. Feedback data: discuss!

9. Components: Cont
Completion – that is at the moment not very well worked out part of the whole system. It requires careful analysis of waters, hydrogen bonding network, multiple conformation. Ligand fitting need to be done here if it has not already been done. For ligands we can use several techniques: For large ligands we can use divide and conqure. I.e. divide ligand into smaller building blocks and try to fit them first and then try to make sense of things. For small ligands we need to generate all possible conformations and try to fit them one after another and score according to experimental data as well as prior knowledge.

10. Components: Cont
Each component should be independent. It means that they can be replaced at any time with another “better” component.
User should be able to enter at any point. For example user can start from interpretation. The system will have to be able to work with minimal information.
Data Base of knowledge is very important component. It needs to be worked on very carefully.