From One-at-a-Time to Automation and Data Modelling in Media Optimisation 🔄

Before diving into the incredible world of a hypergrowth startup like Vow, I spent the previous two decades immersed in researching skeletal muscle growth, development, and repair at The University of Melbourne. One of the final projects I tackled was a deep dive into how an injury (or pathology) leads to an altered metabolic microenvironment. This environment signals the resident stem cell population, directing them to kickstart repair processes. Our approach wasn't about scalpel-like precision; it was more akin to casting a wide net, using whole tissue metabolomics to compare injured and control skeletal muscle. Yet, the patterns we uncovered, especially the differences in the metabolic signature of skeletal muscle during periods of high stem cell growth/proliferation versus differentiation and fusion, were incredibly interesting. They helped us identify hundreds of metabolite targets that might play critical roles in proliferation and differentiation processes (the paper is here for those interested).

Back then, prior to my time working alongside incredible software and automation engineers, our research group methodically tested multiple potential targets one by one in cell cultures. This process, while yielding significant insights, was painstakingly slow. If I could, I’d go back and nudge myself towards the advantages of automation 🤖, high-throughput techniques, and the predictive power of advanced modelling. These tools not only expedite the research process but also enhance the depth and breadth of our findings.

Reflecting on my four years at Vow, I’m struck by the profound shifts not just in my methodologies, but in my mindset. The journey from individual target testing to embracing the vast capabilities of high-tech tools has been both humbling and exhilarating 😌🚀. For those just getting started with media optimisation, I found the following paper by Zhou et al. to be an incredible primer on media optimization through advanced modelling approaches (see paper here). The simplicity and efficacy of the media design approach depicted, especially in Figure 1 (see below), have left a lasting impression on me.

With the cell and gene therapy industry shifting more towards Industry 4.0, focusing on automation, large-scale data analysis and modelling, and AI, I imagine we're going to see an increasing number of third-party solutions to the media design problem. This should have the flow-on effects of increasing yield and reducing costs - for therapies, for food and for any product developed using biomanufacturing 🌐.