All right, good afternoon everyone. As James mentioned, my name is Clay Overturf.

I’m an Automation Engineer at GENO, with a background in mechanical engineering.

I came over to GENO about three and a half years ago, and I’ve been learning the science, which has been really exciting.

In this talk, I’m going to go over a few different things. I’ll start with a high-level overview of GENO and what we do as a company. Then I’ll talk about our R&D workflow and how we’re using Dynamic Devices’ Lynx to accelerate our strain-to-data process.

All right, so at GENOMATICA, our goal is to address the climate crisis by developing scalable, drop-in solutions that drive the greatest reduction in carbon intensity. What does that mean? We’re basically finding ways to make existing products in cleaner, more sustainable ways.

We accomplish this by using renewable resources like plants and sugars, and by ensuring our manufacturing processes are traceable and transparent. This way, we can ensure the products we use have been made sustainably.

We have a few different product platforms at GENO, and the core technology is really similar across them. We’re converting renewable carbon sources—plants and sugars—using engineered microorganisms to produce target molecules used across a range of industries: cosmetics, textiles, automotive—pretty much anything you touch.

By engineering these microorganisms and scaling the manufacturing processes, we can create more renewable products. This makes sustainability part of everyday life, and that’s our goal.

So, what does this look like in action?

It starts with developing microorganisms. We use our GENO Biomanufacturing Technology, which allows us to convert alternative and renewable feedstocks into widely used molecules. These molecules are then dropped into the products we use daily.

Let’s go a bit deeper into how we actually engineer scalable cell factories.

We use a specific blueprint we call our data cycle: Design, Assemble, Test, and Analyze.

We start with modeling and design, using systems bioengineering to guide strain engineering and experimental approaches. That informs the assemble phase, where we apply synthetic biology, engineer enzymes, and integrate genotypes to improve cell performance.

Then comes the test phase—this is the fun part and my job. In engineering, we use robotics and instrumentation, specifically the Lynx, to deploy complex experiments that test the robustness and performance of our strains. This generates a large volume of data. We run multidimensional analyses and feed everything back into the modeling/design phase. Like any iterative design process, each step informs the next, continuously improving product performance.

Engineering cell factories involves a lot of variables. On the cell engineering side, we have a good understanding of how things work. The model starts with the cell uptaking feedstocks—our renewable carbon sources. Various precursors and metabolic pathways are at play, and we can engineer them to produce more of our desired products—Product X, Product Y, etc. All of this impacts performance.

Then there are bioprocess conditions—oxygen levels, pH, temperature—the environment the cell lives in. It’s critical to control these, as they impact downstream process design, cost, and sustainability—our ultimate goal.

How do we use synthetic biology and an information-rich phenotyping platform to advance cell engineering?

When screening, we need to generate a lot of diversity. We use two approaches:

1. Forward Engineering – rational design and assembly of genetic parts to maximize the desired product.

2. Reverse Engineering – generating unknown diversity, screening it under specific conditions, selecting candidates, and identifying beneficial genotypes.

Let me walk you through an example. This is what we call our small-scale technology—a plate-based fermentation workflow.

In this case, we use reverse engineering to improve substrate utilization in a strain. We generate a lot of unknown diversity and screen under process-relevant conditions. At the end, we collect production data.

We’re currently looking at:

• Qs: the rate of substrate utilization

• Qp: the rate of target molecule production

There’s a range of performance, but we focus on the top-left quadrant—high Qs and Qp. That means our strain is thriving in the environment, utilizing the substrate well, and producing a lot of the target molecule. We can select these candidates and integrate beneficial genotypes into the parent strain.

Let’s dig deeper into process conditions.

These include:

• Substrate composition

• Media components

• pH

• Agitation rate

• Temperature

These all impact strain performance and the resulting phenotype.

In the figure on the right, we see how different genotypes produce different growth phenotypes, and how a single genotype can behave differently under different process conditions. Genotype matters, and process conditions matter. So, how do we develop a workflow that screens all these conditions?

We’ve done it in low-throughput ways on the bench—it’s costly and time-consuming. But with our robotic-integrated Lynx, we can now do it in high-throughput.

That’s what I want to focus on.

What does a high-throughput phenotyping platform look like?

At the core is the Dynamic Devices Lynx. We use the LM900 with a 96VVP head. This VBB (Variable Volume Bulkhead) technology allows for independent control of each channel. It’s 12x faster than an 8-channel pipette, and 96x faster than manual pipetting.

We’ve integrated:

• Controlled plate storage

• An incubator shaker for fermentation

• A plate reader for time-course sampling

• Genera scheduling software

• Method Manager for instrument control

• A SCARA robot arm for moving labware

It can get complex—different conditions per plate, many plates per scientist, multiple users, running 24/7.

On top of the physical infrastructure, we’ve built custom barcoding and data handling using Python. This connects to our LIMS, allowing us to push/pull experimental data, update workspaces, and define reservoir assignments.

Lab setup:

• Thermo Fisher Cytomat – fermentation

• Agilent Flex Robot Arm – labware movement

• BioTek Plate Reader – OD/timecourse sampling

• Dynamic Devices Lynx – liquid handling

• Technik Labware Storage, Quick Bio Refill Station

This allows us to run unattended, mostly overnight.

With this platform, we can run advanced multivariate experiments.

There are many variables in screening process conditions. Using the Lynx, we can generate and control different variables across wells in a 96-well plate. For example:

• Yellow dye = substrate concentration

• Blue dye = media components

• Red dye = selective agents

One single liquid handling event can prepare 96 distinct conditions in under 60 seconds. That’s a game-changer.

We pair that plate with our high-throughput phenotyping platform, using the Lynx to sample cultures at defined intervals. We measure biomass or cell growth, which are key markers of strain health. Combined with end-point production data, we can understand how genotype and process conditions affect strain performance.

Ultimately, we’re mimicking batch fermentation in a 96-well plate.

Let’s go over a couple of examples:

How can we use multivariate assays to inform scaled bioprocesses?

That’s our end goal—scaling from milliliters to 100,000-liter commercial fermenters.

One key question: Do strain improvements seen under small, uniform conditions translate to large-scale fermenters, where cells experience heterogeneous conditions?

These might result from:

• Insufficient mixing

• Substrate feed gradients

• Temperature inconsistencies

These differences matter. Strain performance in small-scale tests doesn’t always predict performance at scale. But by simulating these variables using multivariate assays, we can build a clearer picture of strain performance under a range of conditions.