Menu
The biofoundries and biomanufacturing sites of the future must be deeply integrated with digitalized workflows to maximize productivity and execute clean, sustainable operations.
To unlock the full potential of big data and advanced robotics, life science companies must complement their human brain trust with artificial intelligence (AI). Simply put, the commercial success of novel and complex bioprocesses often hinges on repeatedly sourcing highly technical efficiencies. Given the pace and competitive nature of the life sciences, it’s no longer enough to rely on human intelligence alone. By employing the rapid rise and improved accessibility of machine learning (ML), the bioeconomy can blossom.
To this end, Provectus Algae is developing and harnessing machine learning biotechnology to enhance the speed and efficacy of our biomanufacturing platform. We design our software solutions to integrate with our in-house hardware, opening the door to automated bioprocesses, high-throughput biodiscovery, and cloud-controlled facilities for our partners and clients.
Provectus Algae combines cloud computing, machine learning, and cyber physical manufacturing systems for real-time cellular analytics, bioprocess development, and bioreactor optimization. Importantly, we emphasize machine learning scalability in the development of our AI solutions. In time, we aim to leverage our AI approaches to build fully autonomous biodiscovery workflows and biomanufacturing sites.
Within, we highlight some critical machine learning biotechnology applications in our platform while sharing our vision for the future of bioproduction at Provectus Algae.
To work with livings systems, you must understand how environmental conditions impact them. This holds true even for single-celled organisms used in biomanufacturing, only on a smaller scale. By closely tracking cellular phenotypes and product expression in response to varying media, bioprocess conditions, and genetic backdrops, biomanufacturers can identify ideal cell systems and determine optimal growth conditions to maximize productivity. However, it remains difficult to track and collect a massive amount of phenomic data in real-time such that the information can quickly provide actionable bioprocess insights. This is especially true when relying on manual workflows. Adding to the challenge, microalgae species are incredibly diverse with respect to their cellular structures, life cycles, and phenotypes.
Provectus Algae applies machine learning to execute bioprocess cellular analytics, which we use to inform functional phenomics studies on algal cultures to solve this challenge for microalgae biomanufacturing. With our AI-powered cellular analytics and functional phenomics platform, we can automatically monitor microalgae cultures during bioprocess development and production runs to quickly optimize cellular productivity.
Provectus Algae applies machine learning to execute bioprocess cellular analytics, which we use to inform functional phenomics studies on algal cultures to solve this challenge for microalgae biomanufacturing. With our AI-powered cellular analytics and functional phenomics platform, we can automatically monitor microalgae cultures during bioprocess development and production runs to quickly optimize cellular productivity.
Computational microscopy and computer vision are core to our cellular analytics and functional phenomics platform. Our computational microscope can independently capture images of microalgae cultures and upload big data collections to our cloud architecture. There, our computer vision algorithms inspect and analyze the resulting microscopy images. Once trained on a unique microalgae species, computer vision algorithms can capture key culture characteristics, like cell counts, sexual stages, color, shape, and contamination events. In addition, we can use machine learning to connect computer vision observed characteristics to algae genotypes, bioreactor sensor data, light recipes, and culture conditions, providing detailed functional phenomics information and greater throughput. Users can then remotely access the resulting phenotypic information to make critical bioprocess development decisions.
In addition, we are actively integrating our AI software with our photobioreactors and automated systems to digitalize our workflows further. In time, we will be able to use advanced robotic arms and tracks to automatically collect and deliver algae culture samples to our computational microscope. Ultimately, we plan to build these tools into our biofoundry so that we can perform parallel cellular analytics and functional phenomics on many algae cultures. Furthermore, we intend to use these integrated machine learning and automated workflows to enable real-time automated bioprocess data analytics of manufacturing runs. Once online, real-time bioprocess monitoring will allow us to ensure the smooth execution of pilot and large-scale production runs while avoiding costly failures.
To harness a variety of diverse microalgae for their unique metabolites and biomanufacturing capabilities, we must build bioprocesses and photobioreactors that suit the unique nature of each species. Though it’s tempting to assume microalgae species behave the same in bioproduction systems, differences in preferred growth conditions and cellular fragility necessitate bioprocess bioreactor optimization to maximize each species’ biomanufacturing productivity.
For photosynthetic microalgae biomanufacturing, determining the ideal light conditions for expressing a target bioproduct is the most important bioprocess development step. Available light is the primary stimulus that photosynthetic microalgae must react to in their environmental niches. Thus, microalgae species evolved highly specialized complex collections of photoreceptors and photosystems responsible for tuning their gene expression. So, if we want to capitalize on the unique biomanufacturing capabilities of novel microalgae species, we must be able to understand this relationship to optimize light recipes.
Our Precision Photosynthesis™ technology lets us tightly control light parameters and rapidly determine ideal light recipes for maximizing biomass and cellular productivity. Our Precision Photosynthesis workflow uses reinforcement learning (RL) to help us screen and evaluate light conditions. Using Precision Photosynthesis and RL in our biofoundry, we can quickly converge on ideal light recipes for a specific bioprocess, even for unexplored species.
Going further, it’s also clear that hardware can impact the generation of specific products, even if algae growth rates remain relatively constant. We can again turn to machine learning biotechnology to continuously improve, iterate, and adapt our algae bioreactor designs for specific applications. Provectus Algae also aims to develop ML tools to make microalgae species (or bioproduct) specific optimizations to our algae bioreactor designs.
In time, we will also explore the application of reinforcement learning and evolutional strategy learning to model and predict how design changes will impact the bioproduction of specific algae or molecules. Importantly, each production run can help supply more data to these systems for further training. As more data is collected, these ML tools will be able to iteratively improve their predictive capabilities, which we can use to create new custom hardware that improves bioprocess scalability. While this application is not fully formed, it will eventually support better predictability in algae bioreactor design to ensure more seamless scale-up of novel species and products.
The long-term vision of Provectus Algae is to build autonomous biodiscovery and biomanufacturing capabilities for our partners and clients. Naturally, AI and machine learning biotechnology will play an enormous role in achieving that vision.
As part of this vision, we aim to transition our biodiscovery platform towards being increasingly AI-led. Ultimately, we aim to have AI parse through the big data found in our microalgae library and omics database to identify candidate microalgae and molecules that suit the specific needs of our partners. Natural language processing and AI tools that can deconvolute complex chemical spectra to identify metabolites of interest will play a role in this conversion.
Furthermore, we also aim to create manufacturing sites that can be controlled remotely and operated autonomously, building off of our cellular analytics, functional phenomics platform, and existing automated workflows. Ideally, a bioprocess can be automatically run from start to finish, with minimal human intervention. In doing so, personnel at these sites can devote their time to the most complex tasks that require human creativity and ingenuity. While these remain long-term goals for the company, continued development of cyber physical manufacturing systems will make way for next-gen biodiscovery and bioproduction systems that further increase the efficiency and sustainability of our operations.
Provectus IP Pty Ltd is the owner of the following trade marks: PROVECTUS ALGAE; PROVECTUS ALGAE & LOGO; ALGAE CELL LOGO; NATURE, ACCELERATED.; NATURE, ACCELERATED. & LOGO; PRECISION PHOTOSYNTHESIS.