First off, I’m not sure if this topic really belongs in Schedulers or if it would fit better in LIMS & Systems Integrations or if it deserves a new category. Schedulers seems to be the closest fit to me, so here we are.
I have been hearing more and more buzz about “Orchestration” software that enables some really powerful features by integrating on top of the robot scheduling software. On my systems, we do not use any orchestration software and I want to learn more about what products are available, what other people are using, and what is possible with orchestration that is not possible without it. I am hoping to educate myself a little more on what is available and where the value lies.
I have spoken to Artificial and their product seems really flexible. I would love to hear about how customers are using it. Ganymede provides a different but somewhat overlapping option and may not really be an orchestration software, per se. Any Ganymede users want to comment on your experience? I have heard about GBG Orchestrator, but have not researched it in depth - are there any users who want to comment? What other products am I missing? What is the most compelling thing orchestration provides to your lab?
Maybe @DerekDulek has a better sales pitch but my understanding with Orchestrators is that they give a lot more “out-of-the-box” functionality on top of the scheduler which means that a team can spend less time building those platforms.
With larger teams, they may not need Orchestrator software because they can afford to expend resources to build similar tools or already have tools built out. On the flip side, you may want your team to focus more on the science (or your ENG team is maxed out) so Orchestrator software can perform some of that mundane heavy lifting for you like building dashboards, capturing inventory or IoT data.
IMO the best things they offer are
- Instrument Agnostic UI + Data Capture
- Dashboards for the Data
- Infrastructure / Data Services (Most important for me)
- API’s for those data services
Furthermore once you’ve opened up the Event/Actions, Ordering world, there’s a lot there that can be accomplished that’s not unlike what DevOps can do.
Hello! I guess I can chime in a little bit since I’ve been playing a little bit with Biosero’s Orchestrator lately. Full disclosure: I currently just started working at Biosero as an FAS so I may be biased. @DerekDulek is one of our sales people that also contributes to the forum a bit. If you’d like to schedule a tour of our acceleration lab and see GBG and Orchestrator in action it’ll give you more insight as to what’s going on in the background.
Basically we like to unofficially call Orchestrator the scheduler of schedulers, but its so much more than that. The basic functionality from it is to be able to control multiple workcells that would typically be islands of automation not talking to each other. This can be done as simple as a shuttle system, mobile robot, or even just a human picking up and placing plates from one system to another. The real power with Orchestration though, is when you start pairing your whole lab’s infrastructure into it. That means getting your workcells to not just talk amongst each other running experiments, but integrating your data services like raw data processing, LIMs, ELN, and lab data streams.
I’ll leave Derek to talk more in detail about it, but as the lab of the future continues to grow so does the IT side of things. Labs now should be thinking about how to get all that stuff to talk to each other and spit out meaningful information so that we can do science more efficiently. Orchestrator should be poised to do all of the above.
Data services seem/ pretty dang cool, I wish I could see better documentation on it but it’s new so I get that it’s still under wraps a bit.
So when you are talking about data capture and dashboards, do you mean experimental results or data related to system performance that might be more metadata for the experimental results? What are some of the examples of data services that are possible?
As far as I know, data capture and dashboards can capture and display any data that is relevant to the user. Orchestrator should be setup to allow any data stream a user deems relevant to their situation. It can be experimental results (raw or processed), system performance data, run times, system utilization, etc. So long as the data is in a format capturable (raw input/output files) or a 3rd party system with an API, you should be able to integrate those streams. I’m still wrapping my head around all this as well, so if you need a more accurate answer I can get you ahold of someone more knowledgeable about Orchestrator and it’s actual capabilities.
Thanks everyone; I really appreciate this discussion. It sounds like there are some helpful things that can be done with data in orchestration software. How would the data integration of an orchestrator compare to something like Tetrascience Data Cloud?
Well that’s the rub right? What’s going to separate one orchestrator company or tool from the next is what they’re willing/able to provide out of the box or help you create. This variability is what’s going to vary from company to company so you’ll need to do some homework on your needs, costs and what tool(s) make the most sense for your workflow. This is where partnerships can really help these companies carve out a more useful tool set.
With that said, some of the boundaries of what constitutes A or B when it comes to SaaS tools are kind of ambiguous and even though the number of tools available are reaching a bit of critical mass, there’s still no single player that I think owns each space.
As a testament to the aforementioned, Jay Rughani highlighted a few of the companies in the space. As you can tell from a cursory glance, some of the functionality overlaps. To further complicate the issues, these companies may NEVER describe themselves as Box A or Box B (despite being used that way) due to external business pressures.
@luisvillaautomata, @ehung and @DCurrier
Been a bit behind on keeping up with this forum as of late. Apologies, but when I think of lab orchestration, I look at it as a big picture. When we schedule a workflow on a closed integrated system, the focus of the scheduler is largely managing the devices tied to that system. With Orchestration, we remove the boundaries from the workcell to encompass entire labs, facilities, or even global infrastructure.
At the heart of science is data, and what GBG Orchestrator allows is that data to flow and be managed. This can include data generated from an integrated workcell, this could be data from stand alone instruments, or data captured as part of day-to-day laboratory operations (Temperature/Humidity, reagent creation, etc). While I would say that labs are generally pretty good at tracking processes, Orchestrator allows that data to be captured digitally, and moved and displayed as needed.
@luisvillaautomata brings up some great points, that finding the right partner is key in this, as lab orchestration software is not a one size fits all type of situation. The goal at Biosero at least is to provide the base tools that can be deployed and tailored to each lab as needed. Under the GBG Orchestrator Umbrella:
- Data Services-Cloud based application that acts as the central hub connecting all pieces of the overall architecture.
- Insights: Graphical interface that visualizes processes, real time feedback from integrated workcells/instruments
- Scheduler: Schedules and executes processes on integrated platforms
- Workflow Conductor: Easily build workflows that can span across different integrated workcells and standalone instrumentation.
- Workflow Conductor: Plan and direct workflows. Simply stated where workflows built in conductor can be executed.
- Transportation Manager: Materials need to move, whether human, track, conveyer, XPlaner, or mobile robot. This tool helps manage the information around labware moving across the lab.
- LabExperience: A tool to help integrate manual processes and tasks into the digital environment (vial weighing, reagent prep, options are endless)
- IOT Agent: Manage standalone equipment and processes.
The tools are here to use if needed, but not all deployments or orchestrated environments require them all. Happy to discuss this further with anyone interested, drop me a message on here and we can connect to discuss further.