Your Chromatography Data, Speaking One Language

Hello, everyone. Thank you very much for joining us today. My name is Ryan Bernhardt. I'm the general manager of Vercidian and one of the leaders of our Luma platform. I'm joined today with Bill Farrell, senior principal application scientist. Bill, would you like to say hello? Yes. Hello. I'm a senior application scientist here at Residian in the Dotmatics family of applications. And I've recently come off thirty five years of being in the laboratory and being in pharma. So I have a unique expertise in this area. Thank you so much for joining us, Bill. Today, we're excited to talk with you about, our Luma instrument harmonization and more specifically chromatography harmonization. We're gonna be talking about how your chromatography data speaking one language and how Dotmatics LUMA brings chromatography data from every vendor in CDS into one connected AI ready view without having to touch each of those instruments. So I wanted to first start by just kind of orienting, all of us to the laboratory space. Many of you today have experienced exactly what the image here on the right hand side of the screen is referring to. We have a highly technical laboratory space where cutting edge science is being done, but so oftentimes we're faced with a lot of very sophisticated systems, but those systems are disconnected. They're disparate. We have independent instrumentation that is isolated from one another where maybe the liquid handler doesn't know anything about the the samples or the results that the LCMS sitting next to it is producing. And we also have separated expertise in different labs and different areas of the research and development that we're working on. And so, it really creates a difficult and inefficient process of being able to stitch these different disparate systems together. And so if we're serious about streamlining scientific innovation, we need a way to manage every aspect of the lab or better yet orchestrate it. And so as part of this definition of orchestration, it's really about being able to bring the digital, the physical, and the logical elements of the laboratory space together in a harmonized way. However, arriving at the orchestrated lab, is is a journey. It can be a crawl, walk, run, sprint journey, and and and we refer to this journey as a laboratory's digital transformation journey, where you don't necessarily get to the orchestrated lab overnight. In fact, we refer to the laboratory digital transformation journey, as as really a five step or five phases, of achieving the orchestrated lab. At the bottom, it's really the fragmented laboratory space. And this is this is a lab where, there may be systems, in place such as an ELN, but there's, they're completely separated or fragmented, and it relies heavily on the human scientist to make those connections. So we may be copying and pasting data from a result file into an Excel spreadsheet. We may be entering information at the instrument interface to to to get sample list ready for running. And then from there, you may move into what we call the partially connected laboratory space. And this is where we begin to to create some connectivity where maybe we've we've written a macro to go in and and grab data or results from, from a result file and and populate it into a spreadsheet, or maybe we've started to make a connection where a spreadsheet is is automatically generated and and and dumped into a a directory where an instrument is automatically picking that up. And then as we move more up this the the the digital transformation path, we we get into the structured phase. And the structured phase is really where we begin to start thinking about how do we how do we, connect data that is in a common more common format or in a structured way. So we begin to make, connections where data becomes usable across different platforms. And then the harmonized phase is really about being able to bring and and harmonize, all of this disparate data together in more of a unified and and and, single, platform where we can interact with it. And the final phase is really achieving the orchestrated and AI ready laboratory. And this is about being able to to connect the digital, the physical, and the logical elements together in a way where we have data that's structured and and in a fair format that is AI ready or AI accessible such that we can really reduce the the the level of friction in that laboratory today. And so our LUMA platform enables this sort of connectivity that allows us to be able to connect different digital systems and scientific capabilities together in a way that allows us to then be able to capture data, harmonize that data as part of the part of this platform. In addition to being an a connectivity hub, LUMA is built on a Databricks foundation. So it allows us to to, capture and store large amounts of data, as part of that data lake. And then we have the ability to to make that data available to not only other digital systems, but we also have the ability to pass that data down to instrumentation, or processes that are actually being executed in the laboratory. And in addition to our own portfolio of products, maybe more importantly is the ability to connect this data to the tools and and the scientific systems that you're using in your laboratory, which is really what the plus refers to on the on the bottom of the circle there. So Illumina is, not only a connectivity hub, but it's a scientific intelligence platform. And and and what it allows us to do is create this operating system for science. And as part of that, not only is it an open extensible ecosystem where we can connect into those different digital systems and and and physical instruments and work cells, but it it also gives us core functionality where we have the ability to visualize results and create user experiences and dashboards. We can embed business rules. We have an AI powered foundation, so we can leverage AI in our, LUMA agent, to assist as a coscientist, but it also, allows us to make those connections with the physical realm in the laboratory, and so being able to connect to instruments, work cells, proprietary file formats and types, and allows us to be able to make those connections, capture, and aggregate or harmonize that data together into LUMA. And this is what we refer to as LUMA instrument harmonization. And so this image on the slide here is an image that many of you who've ever walked into a laboratory have seen. This laboratory looks like many laboratories, maybe a little cleaner, but the, the idea is that you walk into this, laboratory and you're you see a variety of different types of instruments, automated liquid handlers, flow cytometers, mass specs, HPLCs, microscopes, high content imagers. We have multimode readers and and, physical property analyzers, thermocyclers. And each one of these instruments made by manufacturer, has different, result formats and and and data outputs, some of which are in files, some are going to database systems, some are being stored, on premise locally while others are are being sent to the cloud. And the interaction of this data is is is historically been manual, and it's required a scientist to go interact and and and go grab data files from different directories or different data systems to be able to then interact and do something more. In addition to the instruments, there's a ton of different data reduction tools and and analysis tools that we have come to to count on or rely on in the laboratory for to to to really get the results in a format that are most useful to impact our portfolio. But the reality is it doesn't stop with this one laboratory. This one laboratory is just one of many laboratories that would be found on any research and development campus, and each one of those laboratories have hens to hundreds of different instruments producing a ton of data and and and generating results, each of which we want to be able to interact with and and and analyze and store as scientists. What this looks like in reality in terms of how we've worked with one large pharma organization was we have the ability to connect over three thousand different instruments across the research campus. As part of the three thousand connected instruments, we're capturing over five terabytes of data per day, which is over five hundred thousand files that are being generated and captured on a daily basis. And this equates to over a billion billions of scientific data points that we're able to capture, transform, and harmonize together as part of the LUMA platform. And we're able to do this because we've we've built our LUMA platform on a Databricks foundation that allows us to to store and govern and interact with really, really large datasets. But one particular subset use case of our LUMA instrument harmonization is what we refer to as LUMA chromatography harmonization. So much like the image that we looked at earlier, of a normal laboratory, you'll also find within almost every laboratory, you'll find chromatography producing instruments. Some laboratories, analytical laboratories, will have a larger population of those chromatography producing instruments. And much like we we already, discussed about the reality of any one lab being just one of many on a, campus, that that's the same reality for chromatography producing instruments. We have, the chromatography instruments in in a single laboratory, but we also have several of these instruments that would be across any given, research and development campus. And many organizations don't just have one campus, but they have multiple campuses in many places around the world. And so LUMA chromatography harmonization is about being able to capture data that is coming off a variety of different chromatography, types of instruments from different manufacturers, vendors, brands, and that produce data in different formats and file types and store data in manufacturing specific CDS systems. And so this in in addition to, this, we actually have, the ability to handle this across many different, application types, market segments, different phases of the business, whether that's small molecule, large molecule. And so that, I'll turn it over to, to, Bill to to discuss the rest. Thank you, Ryan. As Ryan said, the chromatography space can be very fragmented. You want to utilize a specific type of instrument for a specific application. There might be some advantages to a particular mass spectrometer or there might be advantages to the chromatography system. And so you want to be able to focus your efforts on gathering the correct information from the instrumentation that you think does the best job. And so what does that create? That creates an environment where you have this multiple vendor collection instrumentation, and each one of them has their own specific file format. They have their own specific way of, we call it a workflow, you know, sort of how you generate the data, how you input the information that you need to gather. And so each one of those has a nuance to it. And so as you build through your organization, you have this conundrum of collecting this information from a variety of different places, they don't necessarily talk to each other. Compound that with some of the systems are essentially databases themselves. These chromatography systems like Empower and OpenLab and and Chomelian, these are systems that they they literally have databases behind them. And so then the information isn't necessarily readily accessible. And we've come to work with the vendors and utilizing their APIs to pull this information out so that we can produce data in our platform primarily. And then what that does is it actually provides a common data file format and a common data file structure, at which point we can then take this data file structure and we can put it into a product like Luma that can break down the silos of each of the data sources and the application types and unify it into one platform, one format, so that it can be actually utilized in mind, and it can gather the information, and you could do a lot of different things with it as we go on. Let me give you a sense of of essentially what this could look like in a laboratory environment. So, Ryan, if you would move to the next slide, please. So let's look at a ninety six well play. It's a ubiquitous type format, but it it gives you a sense of if we have ninety six wells of samples, that's ninety six samples that I'm going to run on one particular instrument. And so this is the run. So next slide, please. As we go through, you might have in your laboratory, you might have a dozen or or more instrumentation, each one running a plate a day. That's going to give you thousands of samples, thousands of data re results to look at. If we start to expand that document instrumentation to beyond one vendor, let's say we go with a couple of different vendors, each one has a chromatography data system, Now we're talking at two thousand five hundred samples per day. And if we go even further and we start looking across the organization, you might have a bunch of different laboratories within your facility, within your campus, and now you're up around twenty five thousand samples per day. All of this data coming in, all of it in a different format, if we add any more vendors, we just complicate the situation even more. And so what happens is this all falls upon the data, the actually, the analytical scientists. They are the ones who have to manage the data, move it around, generate the reports, collect the data, store it, report it, and this becomes quite a task. And this is where a lot of data analytical scientists spend their time, is really in the reporting. The actual acquisition of the data is really rather fixed. It's essentially the run time times the number of samples, but the data processing and the data reporting and the checking and all of that information, that becomes where the real time sync occurs. And so what do we do with this, by harmonizing this data across your entire fleet? What we can do is connect the chromatography systems and the data systems and any other system for that matter across the vendors, across the the different formats, and take care of this this this problem of disparity between the different formats and harmonize it and make it one format. We can then ingest this into a system where it's fair aligned, so it's findable, accessible, interoperable, reusable. You can utilize this information across this, and now you can do the reporting all from one simple format. I remember when I got into the high throughput arena, we had rather standardized on one vendor because we really didn't want to deal with all the data analysis and actually have to deal with this problem of multiple vendors and multiple reporting systems and so on and so forth. So having this today, this actually takes care of that and allows you to choose the vendors that you want to because the data will become harmonized and centralized in one location. So what does Luma do? It basically pulls this information in, it allows you to utilize it, and it also empowers you to go a little bit further and connect it to other data sources and other information that's coming in. What it does not do is we're not replacing the instruments. We're not replacing the control of the instruments. We're not replacing the CDS of the instrument. Matter of fact, we work very closely with our partners to utilize their APIs and pull the information out just as they provide it, provide the tools to do that for us. And so what we wanna do is focus you on the actual collection of information, collection acquisition, doing that, but take and put this file management in a different location so that you're not dealing with it, that becomes an integrated process, and you can now focus on the science. And so looking at, you know, what it would look like without harmonization, you might have the different, various different data systems, CDS systems, to which you have individual silos of data that you have to go and extract out, and maybe you have to come up with a tool to do that. What we've done with Luma is we actually have connectors that we can pull the information out, and now once it's processed, we can move this information into the Luma environment. So now you have all your all your vendors, all the daters, parsed into one format on one platform and enabled with AI, agent agentic AI to actually pull this information out. And so for, you know, specifically, you know, picking on a a few of the chromatography data systems, looking at Empower and OpenLab and ProMillion and any others that come along, Luma will harmonize this information, bring it in, harmonize it, and then from that Luma platform, you can generate reports and various outputs and even connect to your own internal systems, and that's sort of like the benefit of having one platform is it makes all of this a little simpler. I think I'm passing it over to you, Ryan. Thanks, Phil. So in addition to being able to harmonize, all of the data together in that one platform, one format, you now have the ability, once that data is harmonized together, to be able to make connections with other systems to really go beyond harmonization into more of the orchestrated lab if we refer back to that laboratory's digitalization transformation journey. And so what does that actually look like? Well, with harmonized data, you have the ability to start to look at trending across the fleet of chromatography producing instruments. You have visualization of all of that data without having to go and export different formats and and stitch something together. You have cross vendor, cross site comparison. So as Bill mentioned, oftentimes, laboratories have standardized on one particular manufacturer type because they become experts in that. But if you have multiple labs in multiple areas, usually, there there's not a lot of crossover. So being able to pull all of that data together, becomes really valuable from from an analytics and a visibility and and and usability standpoint. And then as part of this, you have a complete historical data set now to go back to to apply AI on, and you have a traceable record, an audit trail that that you also have as part of that for compliance and governance and and and traceability as well. In addition to being able to to leverage this harmonized data, you can now make your data available to your ELN, your LIMS, your asset management tool if you're looking at, you know, instrument utilization and downtime, uptime, performance over time. And so you have the ability to not only connect these systems, but you also have the ability to connect this data as part of complete end to end scientific workflows. And so as part of going back to the the the lab's digital transformation journey, you don't necessarily arrive at the orchestrated lab overnight. And what it looks like in this scenario of chromatography harmonization is that it's originally bringing in chromatography instruments. You started with files and results and and the data coming off in those different formats. Then there is the chromatography data system that many labs have started to leverage, where you have those results now not going into directories, but going into this CDS system. And then the harmonized solution is really being able to bring all of that data from the various CDS systems, others that may produce files, bringing that all together into a single platform, which Luma enables. And then finally, it's about making the connectivity to other systems, other instruments, and work cells so that we can really create the orchestrated lab where we have the ability to apply AI for enabling that design, make, test, analyze cycle. And so what does this actually look like within Luma? And here's an example of a Luma dashboard dashboard or user experience where we're bringing in chromatography data from, in this case, it's the Empower and Chameleon CDSs. You'll see that from one platform interface, we're able to interact with our chromatography data. We have now that data connected to our ELN and chemical registration tools where you can see the identity of the molecule. You can see things like purity and amounts. You can interact with the peaks to look more specifically or drill down into those, but you have that connectivity with your other scientific systems and tools that are allowing you to really create this frictionless environment of being able to to get answers to questions quicker and more effectively than ever before. And so I'll go back to where we started, and that is that if we're serious about streamlining scientific innovation, we need a better way to manage every aspect of the lab or, better yet, orchestrate it. And so LUMA is a platform that is enabling the orchestration of the lab and accelerating scientific discovery in a way that hasn't been possible before. And so chromatography of harmonization is one of those steps along the way to being able to achieve more of an orchestrated lab. And and with that, I wanna thank you all for attending, and we'll open it up for some q and a. I have, I see that there's, there's one one question. I'll read it off. Is it possible to connect the chromatography system with liquid handling system to transfer the plates coming out of chromatography system and get the liquid handling system to perform sample pooling based on the chromatographic data? And I think, you know, it's an interesting question. It's more of an orchestration, right? And this is truly what we are trying to achieve. We work with partners like High Res Bio and Biocereo, and they utilize these robotic systems to do a lot of the transferring. And so we work with them to provide the information to choose, if you would, the pooling and what fractions get sampled and which fractions go forward, and all of that information is actually captured in this Luma platform. And part of what we've been working on since we joined the Dotmatics family is pulling this type of orchestration into the Luma platform. And that's where we are today, is being able to enable this orchestration from this platform as opposed to from a desktop or or a a work cell platform. Yeah. That answers your question. Just to expand on that. Great question. And and I just would highlight that that is a great use case for more of the orchestrated lab. So in the image here on the screen, you can imagine, you know, pulling that data from a CDS. LUMA is able to do the the churn based upon a threshold of, say, if you were you're doing, cherry picking or if you're based upon the purification, you're putting together a volume list for the liquid handler, then we have the ability to essentially create that list, whether that's cherry pick list that's in a file format or whether we're making that list virtually to pass through an API, then we can actually pass that down to the liquid handler for processing. So that's a great that's a great example of an orchestration use case of being able to bring in data from one system and then make that data available to another system for an action. Any other question that have come in? Yeah. The next question, harmonization. Does that mean that even the processing information, e. G, the shape of the baseline, including preview exactly the way it looked in the CDS is possible, or is it more of the raw spectra? The answer is, it's really both. We're working with our partners to bring in the information so that we have both the raw, if you would, X, Y data, if that's the appropriate term for both the chromatograms as well as the mass spectra, but also to bring in that pre processed information that was in the CDS. If it's provided and you've done the processing, we can pull that information in as well. Yeah. And that's that's really the the focus of the harmonization offering is really about being able to bring the the raw and bring, or and or bring in that CDS process data. If if you wanted to go beyond that where you actually would have, the processing of results based upon a certain analytical workflow, that is actually more of Versidian's core business and what we do to actually process the data based upon your analytical application, such as reaction screening or purification or compound QC. But in the case of Luma harmonization, it's really about just being able to capture that data, transform it into a standardized format that's AI accessible, and bring it into to Luma where then we can we can visualize it, we can make it available to other systems, and and and and really about bringing it together into one one place. Any other questions, Bill? Yeah. Yeah. There's one. Are you able to work with analytical data that is first centralized into something that is an SDMS, or are you only able to work with the native data that is generated at the instruments, I. E. Raw files? Great question. So we can integrate with SDMSs as well as capture data from individual instruments. We can also bring in data from other types of systems. So with Luma, we have a number of ingress egress patterns, so we can do file based capture and aggregation. We can do API based. We can do direct database through, you know, SQL and Oracle. So there's a number of different ways that we can interact with systems, and it doesn't have to be data coming directly from an instrument. It could be coming from an ELN, a LIMS, an SDMS. And even there's, you know, historical proprietary data formats that we can also write bioparsers for to be able to kind of unlock that data that maybe was generated fifteen, twenty years ago or is locked away in PDFs. We can extract that data and bring that data together in LUMA alongside data that's been generated in more modern systems. Another question was, what tasks do you use Agenic AI for, and how do you balance the utility of these tools versus the potential risks of data manipulation with constantly evolving and potentially opaque tools? That's a great question because we could probably spend a lot of time on on the answer to that, but I'll I'll give I'll give the as quick an answer as I can. So LUMA has a AI foundation, so an LLM that's that's that's based upon Claude. And then as part of that, we have agentic AI, which we refer to as LUMA agents that act as coscientists. There's there's really three types of modes of of AI use as part of the LUMA platform, one of which would be assistive assistance AI, which is more asking AI to help us with tasks. And and one of those tasks would be like creating a dashboard or user experience or data flow, but it could be other things like you're asking questions about your your your data in terms of, you know, what is the best PKA data that that I've had in the last week. And then and then from that, there'd be more domain specific AI or what we call domain focused AI, and that's actually the using AI that's more specifically tailored and targeted for the scientific capabilities. And as part of the dotmatics portfolio, we have a a rich set of of tools and systems where we have a lot of scientific expertise and data that's been baked in. So you could say, for example, do an auto population of of cells from this flow cytometry experiment, and and our agent Luma could could do that. And then the kind of the holy grail of the AI use would be composite AI. So this is asking asking questions that there's no one mode that of of one mode or model that can actually answer. It's about bringing together a combination of models that are being used based upon the data that's available in your Luma ecosystem as well as data that's available externally. These would be much more scientific rich portfolio predictive capabilities that would be used. And I'll make two more points on that. One is that we have an open and extensible ecosystem, and that's true for our AI use. So Luma, you have the ability, instead of using our Luma AI strategy or models, you can simply point our AI pointer to your own organization's AI strategy, and and Luma will actually leverage your own organization's AI strategy to get to to get to those answers. And then I'll also mention that our AI has been applied in a way that's really geared for science and for scientists, meaning we use deep learning mode on everything. So every answer will have the ability for you to go back and see the assumptions and how our AI models got to that answer, and so that you can actually optimize it so that it's repeatable, and so that you have kind of that full traceability and audit trail, which is really and really important for for scientists and for, you know, the the scientific environment. That looks like, that's all there is, for the questions. Awesome. Well, thank you very much, for attending. And if questions come up, please reach out to the team and happy to dive in to more detail and also explore your needs and how we can enable the orchestrated lab. Thank you very much.