Sure. Omid here. So I think that depends on the customer type. As you remember, the number of customers are relatively small. So the point I'm going to make, it may change over time. So typically, a customer gets the Proteograph, runs a relatively small study, maybe in the tens of samples, to understand the platform, the way it operates. That data set then prompts them to get the resources to run a larger sample. It may be in the tune of many tens of samples or hundreds of samples. And then those samples get run and then the data analysis part of it that follows before they kind of build on. I would say a case in hand would be if you look at the way Oregon Health, for example, played out. They were the first customer to receive the instrument. They ran a relatively small, in the 10s of sample size, study that they then presented that at a conference. That formed the basis for them to secure the resources, which, by the way, is not just funding, it's also access to samples, because these sites are -- they don't necessarily have the number of samples, as you can imagine, ready to go. So they secured the funding and the sample to then run about a 1,000-sample study. That study took approximately just a little bit north of a quarter to complete and now they're in the data analysis part of it. So I would say that's probably representative of a robust customer if you can think about it. Now there are customers that are even more aggressive than that, if you would. A customer like PrognomIQ, obviously, knows the technology well. They hit the ground running with a large number of samples, and those study sizes have only grown and my expectation is they will continue to grow. The other point I would make there related to it is it's also a function of external data being presented. I think that's going to facilitate this for these customers to be able to access resources and also the samples to run. Progressively, we're seeing customers present data. So if you look at the most recent ASMS conference, 17 posters got presented: 10 of them were from Seer; four from PrognomIQ; two from Evotec; one from Cornell. Of that 17, there were four talks: four from PrognomIQ, one from Seer; one from Evotec. We're increasingly seeing customer presentations and customer data that's going to then pave the road for larger and larger studies to get done at a sharper slope.