Comb-e-day e-models Dr Jonathan W Essex University of Southampton.

Comb-e-day

e-models

Dr Jonathan W Essex

University of Southampton

e-models

• Comb-e-chem– End-to-end model:

Experiment

Data Derived results

Pattern searching

Broad conclusionsNCS

Other databasesCalculations

Statistical analysis

Solubility Prediction

• Critical property for the pharmaceutical industry

• Solid-state structure from X-ray diffraction– Structural analysis for patterns (Jeremy Harvey)

• Calculation of single-molecule properties• Calculation of bulk properties• Statistical modelling to yield predicted

solubilities

Solubility Prediction

• Optimisation of the statistical model requires lots of experimental data

• Database– Store known experimental data

• Many values, different sources, different conditions• Some data missing!

• More data arriving all the time

• New types of data arriving

– Store results from different modelling studies– Store and recover workflows– RDF and triple store (Kieron Taylor)

Solubility Prediction

• Calculations– Automatic update of statistical models as new data

become available– How and where are these calculations performed?

• Condor?• United Devices?• Larger grid?

– E-learning (Robert Gledhill & Sarah Kent)– Protein conformational change (condor and small

dedicated cluster) (Christopher Woods)

Investigating conformational change

• Difficult to investigate the conformational change experimentally, as P-NtrC is short-lived

• Simulate the NtrC protein and encourage the conformational change on the computer

Solubility Prediction

• Update existing statistical models using more modern methods– Collaborations between statisticians and chemists– 50 % improvement in model predictions for

solvation free energies (Ralph Mansson)– E-learning (Dan Grove)