The design platform is responsible for the modelling and design of enzyme and regulatory elements, pathway inference and genome-scale modeling. Design implies part/pathway optimization and includes iterative learning.
Enzyme and regulatory element selection: For the in silico design of our synthetic parts we have demonstrated strength in prediction of pathway enzymes (Bioparts mining), sequence feature and reaction rules encoding, 3D modelling and molecular dynamics. DNA part design: selection of DNA blocks, simultaneous codon optimisation and bespoke RBS design; directed evolution design for HTP combinatorial mutagenesis and parts selection for compatibility with multiple assembly methods. we are expanding our existing GeneGenie software for oligomer design into a comprehensive workbench for the design and directed evolution and optimization of enzymes, regulatory elements and entire pathways. The diversity of activities targeted (including novel enzymes) requires accelerated strategies for intelligent Directed Evolution (MD; machine learning; explore large area of the protein sequence/structural search space; BioQSAR). We use machine learning to infer design rules rapidly from large sets of sequences, including the prediction and synthesis of high-performance variants.
Pathway design and selection:Pathway modellingis applied to identify and characterise metabolic pathways leading to chemical diversity. The challenge is rapid identification of optimised relative activities in newly engineered pathways.Building on strengths in dynamic modelling (Copasi) and pathway modelling with incomplete kinetic data (Monte Carlo) and the use of transcriptomics to predict pathway flux (MCA: metabolic control analysis). We are building on our expertise on development and validation of protocols for metabolic pathway design (RetroPath) through the application of a retrosynthetic algorithm that considers putative routes on an extended metabolic space involving pathway enumeration and ranking.
Systems modelling: To characterise the metabolic behaviour and capabilities of the proposed host organism we applySystems modelling expanding our existing metabolic network reconstructions using a combination of transcriptomics, untargeted metabolomics and in silico methods. Development of genome-scale modelling, including flux, thermodynamic and transcriptome-based constraints will improve predictive accuracy beyond standard genome-scale models. We are also actively developing approaches for creating genome-scale kinetic models. The challenge is rapid identification of knock-out and overexpression targets for the generation of optimised chassis strains.
Dr Pablo Carbonell – Pathway design and systems modelling (Pablo.Carbonell@manchester.ac.uk)
Dr Neil Swainston – Parts/Devices level modelling (Neil.Swainston@manchester.ac.uk)