First, a little bit about Editco, we’re relatively new as of this year. We’re actually a fully independent company. We came from kind of separated out from a group called Synthego, who made synthetic guide RNA. We’re the cell business. We do CRISPR-edited cells. Our goal is to basically empower scientists. So we have CRISPR expertise, scientists have experiments that need editing, and we can do, you know, we handle primary cells, iPSCs, immortalized cell lines, all kinds of knockouts, knock-ins.

Kind of, if you want it, there’s probably somebody at our company who knows what’s going on and can give advice or even do the edit for you. So, a little look at our product lines. We’ve got a couple reagent products, gene knockout kits. If you’ve got pesky gene that you want to stop working, arrayed gRNA libraries if you have a whole lot of genes, and then all sorts of cell lines.

What we’re going to talk about today is just those guide RNA libraries.

And these two product lines actually are very similar to each other. So, EditCo uses a technology, a strategy called of multi-guide RNA. So we have a given gene target, we actually designed three different guide RNAs for it. What that does is it basically tries to guarantee larger fragment deletions than just like kind of single indels or anything like that. So it’s really effective at knocking out genes, I think up to 80% typically. And we have designs for the entire human genome, so all 20,000. We do the mouse genome as well.

And these product lines that you can just hop online and buy from our website are kind of enormous. We have 20,000 different gene targets.

If you’re trying to buy the kits, they come in little tubes. We have four different yields. They all come in the same tube and that’s 80,000 different products that you’re able to buy just from that single product type. The arrayed libraries get really complicated because we do 96 and 384-well formats. So, five different yields, four different labware, that’s 94 million+ combinations. We have more gene targets for more specialized purposes that we are bringing online. We have more labwares, more yields, but this is getting close to 100 million different product types that you can buy from our website and that we need to be able to fulfill. This is all served by a inventory that we have of these multi-guide RNA designs.

So they’re all stored in tubes and kept at minus 20. Suspended in water at very specific concentrations, but as you all might know, it’s very hard to get things exact in this world. So our inventory, we have a ton of tubes. We have unique gene targets in every tube. They don’t get used at the same time, so we actually have unique volumes in every tube.

Processes aren’t perfect as well, so there’s unique concentrations in every tube. And there are about 50,000 tubes that we have in storage right now.

So when we deliver these products, we deliver them in terms of yield. You know, 150 picomoles, 1.5 nanomoles.

That’s a different volume from every single tube. Every single tube has a different height, every single tube is going to have to behave a little bit differently, and this is kind of a nightmare. How do you fulfill 100 million different products from 50,000 unique tubes?

Our answer to that is you build a work cell. Try and automate as much as possible.

And at the core of our work cell is the Lynx. We have the 730i with a VVP head. VVP is going to come in very handy for us because like I said, every tube needs its own volume.

So this is a, these were existing product lines for us. This was a new automation challenge that we were tackling at EditCo. And ultimately what we needed from a liquid handler was obviously great precision and accuracy. We want to make sure we’re delivering the amount that we sell.

So if you buy 250 picomoles, we’re giving you that. If you buy 250 picomoles and you want, you know, 384 different gene targets, we want to make sure every one of those is delivered. These need to be fast. We have a lot of orders coming in. We need to make sure that we can fulfill them as quickly as possible.

Obviously, when you’re buying reagents for an experiment, you don’t want the delivery of those reagents gating your ability to get the data that you want at the end of the day, so delivering these as quickly as possible, plating them out as quickly as possible.

With the challenge of our inventory system, independent channels was critical. There’s no way we’re going to be able to get away with like a 96-channel flat transfer.

Liquid level detection (LLD) was also critical because we don’t actually, we track how much is in each volume or in each tube, but we want to make sure that we’re actually treating each tube correctly.

We want to support a wide volume range that all those different yields that we offer go from 150 picomoles up to 10 nanomoles. What that translates to is pretty much three microliters to 400 microliters, all on a single work cell. And ultimately, this is a very complicated problem, so we have a lot of very complicated back-end software to make it happen. And we needed a liquid handler that could take what we have and turn it into a final plate.

So what does this automated fulfillment process look like?

This is a very, very simplified version of what’s going on. The first step is to try and assign the most appropriate tube for each customer tube or well on a plate.

What does most appropriate mean? That was a lot of discussions and figuring out how to make that happen.

Ultimately, sometimes customers want multiple wells on a plate with a single tube or with the single gene targets. We need to be able to kind of fulfill any kind of plate map transfer. We need to make sure that we have enough material in that tube in the first place to fulfill.

And we want to make sure that we’re consuming our inventory in a good way. We don’t want to leave stuff sitting at the bottom of our storage for a long time. We want to make sure that we’re receiving it and sending it out in a timely manner.

The next step is to get those tubes into a rack. We’re using the SPT Labtech compound and Lab2Lab for that.

The arranging of the tubes themselves can save you a lot of headaches at the liquid handling stage. Obviously, if you can get more of a head stamp in, then you’re going to save a lot more time. We don’t want to kind of have them randomly arranged in the rack and then have to do individual or 8-channel transfers for the entire thing.

And the last step is to do all the right liquid transfers at all the right volumes into the customer plate. In this little mock-up, this red gene that we have, we only had two tubes and inventories, so we tried to fulfill the first few. You get a head stamp out of it, and then you have to add a couple of hit picks afterwards to finish out the plate.

This is a very dynamic process. A lot of it comes down to arbitrarily mapping two different plate designs from our inventory tubes to the customer plate. And it works pretty well, I’m happy to say.

Next, I wanted to talk about some of the reasons why we’re really happy with this solution and some of the challenges that we encountered along the development process.

So, in integrating the Lynx into this plate fulfillment process, different tubes get used at different rates, they all have different volumes. We can’t do tracking the liquid surface. We’re trying to fix things. Obviously, with the VVP head, you’re fixed at Z. You can’t have different levels. So, we wanted to make sure we’re not overflowing the tubes that are brand new and full while we’re trying to get the ones down low.

What we ended up coming to as a solution with the help of Dynamic Devices is taking advantage of LLD.

We basically go down to a known height where we know that the tubes aren’t going to overflow. We identify any tubes that have liquid above that height, and we can split all of our worklists based on that. We have a unique handling for the high-field tubes where we do their deliverable transfers first, and then we come back and aspirate off a large amount of the liquid in there to help with that displacement challenge. And then we can go all the way to the bottom of these tubes and get all of the remaining customer transfers.

This is a very cool thing to see. Sadly, the video won’t be available today.

So the other thing that we care about a lot here is that we are handling a lot of customer wells on a given day. I think in the last year, we were kind of peaking at around 2,000 inventory tubes being pulled per day. So in our process, that means we were going up to 8,000 independent customer wells being fulfilled and plated out in a single day, five days a week. That’s a lot of data to track. We want to make sure that we’re not getting dropouts. We’re not seeing anomalies.

Luckily, with the VVP head, everything is being recorded, and we can kind of– it’s a lot of data, but you can parse it if you are thoughtful about it. And we can kind of take whatever our inputs are, get some outputs, make sure that things look correct for what we consider a successful transfer, and then have some confidence as well with material usage on our inventory side so that we know what’s going on. This is a very pretty slide, but it’s actually where the video would happen if I had the video.

We just coincidentally chose all of the pinks and purples similar to Dynamic Devices, but it is nice that the instrument lights up in our company’s colors as well.

Oh, no, this is stuck.

Okay, cool. Well, I want to give some numbers on performance. So, this is a liquid handler. This is a complicated liquid handling problem. The good news is we were able to validate kind of operationally everything that we needed to. So we see from 3 microliters up to 190. We’ve got great CVs, we’ve got solid accuracy.

The Lynx is doing the brunt of the work. There’s a lot of other process around it, so it’s really critical to make sure that the performance that we’re seeing here at the liquid handler stage is as optimal as we can so that all the other processes kind of around it have some room to breathe.

Time-wise, it’s also pretty awesome. This is kind of a broad description of how our process works. I’ve got some times here for how long it took to do this on the Lynx and then how long it took to do this similar kind of series of transfers on pick your favorite 8-channel liquid handler. What you can see is we do great when we’re mostly head-stamping, and that does fall off because we end up doing single-channel hit picking. So that is a downside, but the good news is that I went back through our 2024 data and 97.6% of wells were fulfilled via headstamp. Pretty much always we’re doing faster than a different liquid handler could do, which is great for us.

We’re trying to kind of make as many of these plates as we can every day, and so at some point it quickly becomes untenable to take much longer.

The next look is at the actual process qualification. So, not just is our liquid handler doing the volumes that we want, but are we actually delivering the yields that we want to to our customers?

The bounds get a little looser here because there’s some manual operation processes and other elements here that kind of get all rolled in, but what you see is we are very comfortable with CVs and inaccuracies for all of these transfers. And what’s actually going on here is that we have 96-well transfers of 500 picomoles here, 384-well on the other side. We did these replicates, we actually do two plates per run, and then we can also do this weeks apart, and everything kind of behaves the exact same. So robust, reliable to pretty much everything that we can throw at it.

In summary, our experience, the Lynx has great precision and accuracy. Very fast. It does have a VVP head if you choose to get that, and I do recommend it. It kind of solves this very complicated problem for us of like, how do you handle this many plates, how do you handle this many tubes, and how on earth do you sell 100 million different products from your website?

And luckily, we can. So, a brief thank you to the Dynamic Devices team that helped out with our liquid handling method development, and then the EditCo team as well for all of the process validation that went into it. So, thank you.