1. ORCiD
Multi-Objective Transformation based De Novo
Design of Novel Surfactants
Christos C. Kannas 1
, Warren Read 2 3
, Noel Ruddock 3
, Martyn Fletcher 4
, Tom Jackson 4
,
Robert Stevens 2
, Jerry Winter 3
, Peter Willett 1
and Val J. Gillet 1
1
Information School, University of Sheffield, 2
School of Computer Science, University of
Manchester, 3
Unilever and 4
Cybula Ltd
I. Project Overview
II. Surfactant Molecule
Amphiphilic compound
Hydrophobic (oil soluble) component (tail) (1 or more)
Hydrophilic (water soluble) component (head) (1 or more)
Tail
Head
Surfactant
III. Objectives
Design and implement:
A multi objective evolutionary algorithm
Design novel surfactant molecules
Utilise transformation enumeration
IV. Input Examples - Starting Molecules
V. Input Examples - Transformations
Esterification
Sulphation
Ethoxilation
VI. Multi-Objective Search Tool
Evolutionary Algorithm for De Novo Design
3 Objectives (Cost & 2 Surfactant Properties)
Population (surfactants & non-surfactant)
Surfactants & Non-surfactants archives
Steps involved:
Evolution =⇒ Intelligent Transformation Enumeration
Scoring =⇒ Surfactant Properties Calculator & Production Cost
Calculator
Evaluate Solutions by Pareto ranking
Diverse Selection
VII. Intelligent Transformation Enumeration
Molecules (SMILES) & Transformations (SMIRKS)
Automatic matching of molecules to transformations’ reactant patterns
Multi-core parallel computation
Transformations’ products
VIII. Surfactant Properties Calculator
Calculates a number of surfactant properties
Preprocessing step - Pharmacophore based decomposition
Processing step - Surfactant properties calculation
Postprocessing step - filtering surfactants and non- surfactant molecules
IX. Results
1 Step Transformation 2 Step Transformation
X. Outcome
XI. Future Work
Genetic Programming
Tree representation of solutions
Visualise production history
Intermediate products part of the tree
Many Objective Optimisation (3+ objectives)
Pareto Ranking
Preferability Operator (Guide the selection)
Multiple models (calculators/predictors)
Acknowledgements
http://www.sheffield.ac.uk/is/research/groups/chemoinformatics c.kannas -at- sheffield -dot- ac -dot- uk