Periodic Reporting for period 2 - BIOINDUSTRY 4.0 (RI services to promote deep digitalization of Industrial Biotechnology - towards smart biomanufacturing)
Período documentado: 2024-01-01 hasta 2025-06-30
Industrial Biotechnology (IB) is a Key Enabling Technology for the circular bioeconomy, industrial renewal and European manufacturing autonomy. Its development – favoured by the EU’s biotechnology and biomanufacturing initiative – is central to Europe’s strategy for climate neutrality by 2050. For IB to become a mainstream manufacturing pillar, it must embrace advanced digital technologies and data-driven approaches. Expected gains include shorter R&D timelines, lower costs and more reliable, autonomous production systems.
To contribute to smart biomanufacturing, the EU-funded BIOINDUSTRY 4.0 unites wide expertise, including four European research infrastructures, to develop services that spur R&D and innovation in biotechnology. BIOINDUSTRY 4.0 harnesses AI and data-driven approaches to tackle challenges and add value to future R&D. Specifically, it addresses high-quality data and metadata production and interoperability, new methods to valorise microbial biodiversity in culture collections, simulation and real-time control of bioprocesses, and real-time monitoring of bioreactions.
Aligned with European strategy for industrial competitiveness and sovereignty, the project will accelerate bioprocess development for biomanufacturing and consolidate the role of research infrastructures in Europe’s transition to a sustainable and competitive bioeconomy.
To contribute to smart biomanufacturing, the EU-funded BIOINDUSTRY 4.0 unites wide expertise, including four European research infrastructures, to develop services that spur R&D and innovation in biotechnology. BIOINDUSTRY 4.0 harnesses AI and data-driven approaches to tackle challenges and add value to future R&D. Specifically, it addresses high-quality data and metadata production and interoperability, new methods to valorise microbial biodiversity in culture collections, simulation and real-time control of bioprocesses, and real-time monitoring of bioreactions.
Aligned with European strategy for industrial competitiveness and sovereignty, the project will accelerate bioprocess development for biomanufacturing and consolidate the role of research infrastructures in Europe’s transition to a sustainable and competitive bioeconomy.
Over the first 30 months of BIOINDUSTRY 4.0 progress was achieved across all scientific and technical activities.
Work towards a data fabric, framework and compute environment to support transmission to digital twins advanced. Specifications and minimum information models were established, paving the way for a first prototype. This enables structuring of FAIR (open science-compatible) [meta]data sets, preparing them for AI applications. Additional work tested containerised workflows and real-time streaming. Early work on a biotechnology dataspace was also done, preparing a federated data infrastructure for biotechnology R&D.
Using legacy and newly acquired data, multiscale datasets from several microbial use cases were collected, describing physiological behaviour, omics, metabolic dynamics, and physico-chemical features of bioreactions. All were organised with a standard template and uploaded to the Yoda platform. These datasets now train models and support digital twin development.
Preparatory work also progressed: modelling scaffolds were prepared in different coding languages, bioreactor model libraries established, and algorithms for real-time validation implemented. Early tests show good performance on legacy data. Tools identifying metabolic states and optimal feeding strategies are available to enhance productivity and stability. This first phase paves the way for advanced prototypes at project end.
Alongside modelling, progress on Process Analytical Tools (PAT devices) was fruitful. Development led to trials of an in situ microscope for real-time microbial physiology and a Raman-based device at 775 nm. Work on all devices progressed, enabling further testing. Simultaneously, analysis of PAT-generated data advanced, linking physical and digital environments.
To harness microbial diversity, a new microbial strain data standard was developed and shared via GitHub, catalogues were enriched with datasets from legacy data or text-mining, and a prototype for natural language querying of the strain discovery interface (www.strainsbook.org) was tested; further work is underway. If successful, the DSS will accelerate discovery and democratise access to microbial collections.
Finally, stakeholder consultations through interviews and focus groups were held to ensure tools meet user needs and secure uptake of services under development in BIOINDUSTRY 4.0.
Work towards a data fabric, framework and compute environment to support transmission to digital twins advanced. Specifications and minimum information models were established, paving the way for a first prototype. This enables structuring of FAIR (open science-compatible) [meta]data sets, preparing them for AI applications. Additional work tested containerised workflows and real-time streaming. Early work on a biotechnology dataspace was also done, preparing a federated data infrastructure for biotechnology R&D.
Using legacy and newly acquired data, multiscale datasets from several microbial use cases were collected, describing physiological behaviour, omics, metabolic dynamics, and physico-chemical features of bioreactions. All were organised with a standard template and uploaded to the Yoda platform. These datasets now train models and support digital twin development.
Preparatory work also progressed: modelling scaffolds were prepared in different coding languages, bioreactor model libraries established, and algorithms for real-time validation implemented. Early tests show good performance on legacy data. Tools identifying metabolic states and optimal feeding strategies are available to enhance productivity and stability. This first phase paves the way for advanced prototypes at project end.
Alongside modelling, progress on Process Analytical Tools (PAT devices) was fruitful. Development led to trials of an in situ microscope for real-time microbial physiology and a Raman-based device at 775 nm. Work on all devices progressed, enabling further testing. Simultaneously, analysis of PAT-generated data advanced, linking physical and digital environments.
To harness microbial diversity, a new microbial strain data standard was developed and shared via GitHub, catalogues were enriched with datasets from legacy data or text-mining, and a prototype for natural language querying of the strain discovery interface (www.strainsbook.org) was tested; further work is underway. If successful, the DSS will accelerate discovery and democratise access to microbial collections.
Finally, stakeholder consultations through interviews and focus groups were held to ensure tools meet user needs and secure uptake of services under development in BIOINDUSTRY 4.0.
While the notion of FAIR data is gaining ground, the number of publicly available heterogeneous datasets describing complex bioprocesses remains limited, due to structuring difficulties and confidentiality. BIOINDUSTRY 4.0 addresses this by creating and curating multiscale datasets. A first prototype of a data fabric has been released, showing the feasibility of pipelines sending data to and from digital twins. Once operational, this federated data infrastructure will underpin innovation and enable uptake of smart bioprocess technologies.
At present, advanced information cannot be extracted from microbial databases using multi-criteria natural language queries. In BIOINDUSTRY 4.0 work is underway to make this feasible. A new strain data standard lays the basis for interoperability between microbial collections worldwide. Moreover, development of a prototype shows that natural language querying of microbial databases is possible. As a first step, the Strain Discovery portal already facilitates searches across key collections with enriched data. Long term, this will enable new search capabilities to guide development of microbial bioprocesses.
Currently, digital twins are of interest in many application areas. In biotechnology, the prospect of using them to control biomanufacturing systems is exciting but still unattained, mainly due to system complexity and difficulty of real-time monitoring. Creating a control loop between physical and digital environments requires predictive models, data analytics and novel PAT devices. In this regard, BIOINDUSTRY 4.0 is delivering progress. For example, novel Raman-based and optical prototype instruments show it is possible to access new bioreaction parameters and enrich data for digital twins. Similarly, the project has delivered AI models that reproduce experimental data. Prototype digital twins reveal they can identify metabolic states and optimise bioreactor feeding. Together, these results pave the way to autonomous bioprocess monitoring and control, a step towards smart biomanufacturing.
At present, advanced information cannot be extracted from microbial databases using multi-criteria natural language queries. In BIOINDUSTRY 4.0 work is underway to make this feasible. A new strain data standard lays the basis for interoperability between microbial collections worldwide. Moreover, development of a prototype shows that natural language querying of microbial databases is possible. As a first step, the Strain Discovery portal already facilitates searches across key collections with enriched data. Long term, this will enable new search capabilities to guide development of microbial bioprocesses.
Currently, digital twins are of interest in many application areas. In biotechnology, the prospect of using them to control biomanufacturing systems is exciting but still unattained, mainly due to system complexity and difficulty of real-time monitoring. Creating a control loop between physical and digital environments requires predictive models, data analytics and novel PAT devices. In this regard, BIOINDUSTRY 4.0 is delivering progress. For example, novel Raman-based and optical prototype instruments show it is possible to access new bioreaction parameters and enrich data for digital twins. Similarly, the project has delivered AI models that reproduce experimental data. Prototype digital twins reveal they can identify metabolic states and optimise bioreactor feeding. Together, these results pave the way to autonomous bioprocess monitoring and control, a step towards smart biomanufacturing.