Package Number 1

Title Project management
Starting Month 1
Ending Month 36

Actions Objectives and description of work
The overall aim is to ensure timely and high-quality implementation of the project and to provide timely and efficient administration. The objectives of coordination are to efficiently organize the day to day operation of the project, as well as to establish dissemination, technology implementation plans and strategic management and supervision of the
achievements.
WP1.1. Coordination and day-to-day management (M1-M36)
The task will provide support on administrative, legal and financial aspects. The standard monthly, 6-monthly, and annual reports on progress. Financial management of the consortium includes establishing and maintaining financial records, co-ordination of the cost statement submission and their consolidation, follow-up of payments, and financial reporting. Any deviation is investigated, and remedial actions are taken when needed.
WP1.2. Knowledge management (M1-M36)
The external and internal web sites will be created. Knowledge management and other innovation activities will be coordinated. Publications will be sent to well-acknowledged journals. The implementation plan will be prepared and updated.

Package Number 2

Title Screening and interpreting in-silico Design experiments
Starting Month 1
Ending Month 36

Actions Objectives and description of work
The objective of WP2 would be to systematize the selection and build tools for decision-making using promising technologies for supervised and semi-supervised machine learning. The objectives would be to: (a) consolidate the range of options into smaller sets of knowledge clusters easier to comprehend, inspect, and analyze, (b) produce pathway rankers to (automatically) list and recommend candidates for Build stages. Tasks include
WP2.1 – Data collection and management (M1-M24)
Research will be piloted on bio-hydrogen production from dark fermentation. The work will explore the space of feasible bio-transformations around biohydrogen, butyric acid, and acetic acid to identify potential precursors and possible (de novo) reactions that produce these compounds. Developments will make use of BNICE.ch6,7 (Biochemical Network Integrated Computational Explorer), a computational tool for retro-biosynthesis. They will subsequently data-mine the ATLAS7 of Biochemistry, a catalogue that contains all reactions from the KEGG database and over 130,000 novel enzymatic reactions between KEGG compounds.
WP2.2 – Development of machine learning tools and data analytics (M12-M36)
Machine learning technology will be applied by means of supervised and semi-supervised machine learning. Supervised learning will combine quantitative labels for attributes (e.g. performance related metrics) testing a set of promising norms that describe similarity measures between candidate pathways. Unsupervised learning will explore association rule learning extracting knowledge to use at Build. Motivated by the pilot study, the research aspires to extending the experience for the general problem.

Package Number 3

Title Reconciling theoretical and experimental kinetics
Starting Month 1
Ending Month 36

Actions Objectives and description of work
WP3 will connect Design & Build stages with Test and experiments. WP3 will explore machine learning technology experimenting with logistic regression and classification methods (supervised models) as well as clustering, visualization, and reduction (unsupervised learning) with a purpose to systematically connect in-silico propositions with real-life measurements. Tasks include:
WP3.1 Development of thermodynamically curated models (M1-M20)
Research will make use of high-quality genome-scale metabolic models (GEM) curated and updated with thermodynamic information to allow integration of metabolomics data and estimation of intracellular metabolite concentrations of host organisms using the thermodynamic-based flux analysis (TFA). Such models have been used as a scaffold for building large-scale kinetic models typically consisting of 300+ reactions and 200+ metabolites (including small molecules, protons, and electrons). In the absence of kinetic information (from experiments or the literature), we will use enzyme databases such as BRENDA and SABIO-RK. WP3.2 Advanced analytics to reconcile with experimental kinetics (M12-M36) The technology will be tested on 3 data sets available in the IBISBA consortium. Machine learning technology will explore Kernel PCA and LLE (Local Linear Embedding) for reduction and t-Stochastic Neighbour Embedding (t-SNE) for the visualization of reaction pathways. The research will make use of EPFL data produced using ORACLE (Optimization and Risk Assessment of Complex Living Entities). Kinetic information will be produced and modelled using Monte- Carlo sampling over undefined parameters.

Package Number 4

Title Simultaneous pathway and process intensification
Starting Month 1
Ending Month 36

Actions Objectives and description of work
WP4 systematically connects DBTL with process synthesis and process integration. The objective would be to extend the conventional FBA formulations with reaction-separation superstructures. The purpose would be to produce synthesis technology with capabilities to explore pathways simultaneously addressing engineering efficiency and downstream costs. Tasks include:
WP4.1 Development of superstructure models (M1-M20)
Research will be prototyped on syngas fermentation (Wood–Ljungdahl pathway) capitalizing on previous and ongoing collaboration within IBISBA. Research is intended to discover promising lines for innovations as well as novel configurations that could be validated experimentally. The challenge involves a two-stage optimization problem featuring additional degrees of freedom concerning the chemistries and the chemicals involved. In the first stage the analysis will screen and set up supersets of pathways and metabolomics to include in the superstructure.
WP4.2 Development and evaluation of process intensification schemes (M1-M36)
The second stage will analyze and optimize reaction-separation superstructures featuring options for reactive separations and in-situ product recovery schemes. The reference process produces mixtures of ethanol, butanol, and butyric acid featuring several azeotropes (challenges for distillation) but also several alternatives (stripping, adsorption, solvent extraction) to either replace or complement distillation. The analysis will test the optimization ability to systematically screen against the options. The analysis will also scope for simplifications in the modelling and optimization approach.

Package Number 5

Title Building resistant biochemical systems to perform in industrial environments
Starting Month 1
Ending Month 36

Actions Objectives and description of work
The purpose of the WP5 would be to improve the ability of DBTL cycles to deliver biocatalysts that perform robustly in industrial scale environments where concentrations of reactants are distributed, and mass transfer may hold a significant impact on the overall reaction performance. Research will convert general-purpose models based on flux-based analysis into a family of bespoke models customized for common types of single and multi-phase biochemical reactors.
Tasks include:
WP5.1 Development of models combining metabolomics with reaction engineering (M1-M20) Modelling will combine reaction engineering models with FBA balances considering 2 singlephase reactors and 4 multi-phase reactors. Reactor models will feature options for different distribution functions, mass transfer coefficients, macro- and micro-mixing profiles, and different options to lay out and connect individual reactor compartments. The work builds on multi-phase reactor superstructures available in the group but metabolomics will account for completely new features.
WP5.2 High-throughput testing and analysis (M1-M36)
The mathematical formulations in WP5.1 will be optimized against streams of potential pathways using MINLP programming technology. The analysis will produce pathways that are robust to variations and process uncertainties. It will also point to the types of reactors that perform best. The work will probably rely on mathematical reformulations for the optimization models as the approach should be fast and robust. To the latter WP5.2 will experiment with cutting planes and outer approximation methods such as the ones used in BARON and ANTIGONE.

Package Number 6

TitleExploitation & dissemination
Starting Month 1
Ending Month 36

Actions Objectives and description of work
The WP will exploit the IBISBA infrastructure to promote the work widely in Europe. The objective would be to effectively disseminate and exploit DEBONAIR activities and results among important stakeholders and end users as well as citizens, and the general public. This work package will also include a variety of communication tools to promote actively exchanges between research institutions, stakeholders, and end users. Potentially valuable intellectual property will be captured, evaluated and protected by the appropriate means.
T6.1 Communication, Dissemination and Exploitation Plan (M1-M36)
A detailed plan for communication, dissemination, and exploitation will be elaborated. A draft will be available in M6 and will be maintained and updated every 6 months throughout the lifetime of the project. The final dissemination and exploitation plan will describe actual achievements as well as outline foreseen follow-up exploitation activities in collaboration with potential end users.
T6.2 Exploitation and Intellectual Property Rights (M20-M36)
Pathways towards commercial exploitation will be considered throughout the project in close collaboration with IBISBA. Follow-up initiatives and projects will be considered and conceptualized with interested scientific and private sector partners.
T6.3 Website, development of information materials and electronic newsletter, and stakeholder list (M6-M36)
The project will establish a web presence, will be hosted at IBISBA, and will actively favor ecommunication routes to reach specific stakeholders and end users in scientific and industrial communities, as well as the general public.
T6.4 Publications and presentations (M6-M36)
Activities and results will be presented and disseminated on national and international events. This will include oral and visual presentations, but primarily articles in well-cited, top-quaility magazines and journals.