Advice needed. which system to buy

[TanjaMelbs]

Hi all,
I hope I can pick the brain of some experts here. For my chemistry lab (university) I want to buy a liquid handler for reaction screening. I narrowed it down to the Opentrons OT-2 and the Hamilton Microlab Prep. Pricewise they have come pretty close for us with the recent price hike of Opentrons.

We need for our plans high flexibility, ability to fill 96 well plates in each well with individual mixtures, python control and a system that is easy to extend and integrate into other workflows as good as possible. Despite the rate hike, I still have a tendency to the OT-2. Can anyone help us with some advice and experience? As far as I can see the main difference would be the way the pipette holders work. And potentially how open the python interface is.

Any comment would be great, thanks!

[Kastronaut]

The OT-2 would be flexible in terms of python interfacing for certain, I am uncertain about the python interface with a Prep since I have never used the prep. Hamilton is certainly the tried and true brand in the liquid handler space.

After looking at the Prep software, it may be beneficial to look at the python posts on this forum as one of those could potentially interface with it.

Whichever you do pick, I recommend a service contract and annual PM to ensure your system is always working well.

[UCantBcereus]

I haven’t used the Hamilton MicroLab Prep, so I cant speak directly to that. There are a few things that you should consider/be able to live with when purchasing an OT-2.

First, I’m not sure what kind of chemicals you are planning to pipette. I’d check out the Opentrons chemical resistance chart to see if you are planning to use any chemicals that “poor” compatibility, particularly with the internal pipette plunger O-ring. I had a group that was thinking about getting one for metabolomics work, but was basically told that we would have to replace our pipettes every year due to the solvents they normally use. The Single Channels are currently running $2,250 and the Multichannel are running $3,250. If that’s something you can afford to do in the event of a pipette failure, then it may still make sense. I’d ask the same question about compatibility of your chemicals with Hamilton too.

Second, you need to be willing to spend a lot of time doing labware calibration on the OT-2. especially if you plan to run a bunch of different protocols, rather than a few specific ones many times, or if you are planning on swapping out pipettes on a regular basis. See this link for how to do a deck calibration, and follow along to see the work required to finish various other calibrations (tip length, pipette offset). You then need to Calibrate labware offsets for each labware on the deck for each protocol. If you run only a few protocols over and over, then these offsets will last you a while. If you plan on running a diverse set of protocols, then the 20 run limit of offsets being saved will require you to do this kind of calibration a lot more often. If you plan on swapping pipettes often, you will have to repeat most of these calibrations when you do that. The 8-channel pipettes are also a little finicky to level just right too, expect to play around with that for a little bit until you get a good feel for that.

Third, the API evolves very quickly, to the point that some features are deprecated or are broken, and its not always a fast turnaround or fix. You’ll have to carefully evaluate each software update, to see if the updates are worth it in that situation.

Not all is bad though! Again, with the Opentrons python API, you can be quite flexible with your protocols and is the only “vendor approved” python language. If you can deal with the issues above, it does work quite well. We use it mostly for qPCR in a 384 well prep, and it’s a little workhorse that performs quite well with it. As long as you stay on top of those calibrations, it was pretty reliable to set up 10 uL reactions in our 384 well plates, and do serial dilutions to dilute the libraries before we load them into the plate. And as you note, there is not a lot of instruments at that low of a price point.

From what I understand, the Microlab Prep would much more of a closed box, and if you do need to flexible, that may be a challenge. I’d conceptualize the exact protocols you want to run, and bring it Hamilton and see if it would be possible on the Microlab Prep.

There’s plenty of content on here if you search for OT-2 or Microlab prep, to see what the common challenges are (or the cool thing people have done with either). Like I said, if you are willing to deal with the issues mentioned above for Opentrons, then it sounds like you are leaning that way. I wanted to post this though so that you are at least aware of its weak points, and that you can make an educated decision!

@Kastronaut , Opentrons didn’t offer service contracts until quite recently. I don’t necessarily think that you need one for an OT-2, I’d ask to see what is included as part of these service contract/PM. We do regular maintenance on our OT-2s, which is just repeating all the instrument/pipette/labware calibration quarterly, as well as replacing the external O-rings on the 8-channel pipettes we use. Unless the service contract is very inexpensive for an OT-2, I’m not sure its worth it if you take good care of it yourself. Ours has been going since late 2020 with no major hardware issues to speak of.

[evwolfson]

I’ve used both. To keep it simple…

OT-2 if you desire flexibility and want to develop something truly optimized for your needs. May require a medium time investment to get working, and a large investment to get working well. Requires aptitude in python for advanced scripting, but it’s definitely not required for basic functions.

ML Prep if you want something immediately effective. Only thing I can say is that this thing just works really well and pretty much completes tasks well first time, every time. You can learn the scripting in less than a day, and might be a better teaching tool for introduction/outreach to automation for those less coding-inclined. Support is quite good (props @Michael_Vuke, Clinton, et al).

Both of these instruments have hitpicking so your main ask is fulfilled either way. You’ve narrowed down to two good options so I don’t think you’ll be disappointed either way.

[luisvillaautomata]

The other thing I would add is that Opentrons is about to unveil some very cool AI tools and marketplaces. If you’re going to use these systems with common-ish protocols, that could be critical for building your protocols with speed and so the learning speed with Python programming may end up being severely reduced.

If you like the OT-2 but what some modern tools (like auto calibration) and find more room in your budget, the Flex becomes an option and you can just take those protocols and with minimal adjustments run them on the Opentrons Flex.

[Michael_Vuke]

Thanks for the props, Evan!

@TanjaMelbs I’m not going to go into full sales rep mode here–this is a forum for getting public feedback. But I will say that there are some key differences in technology [Prep uses the same pipetting tech as the Star system–I can unbiasedly say it is the best pipetting tech on the market at this price point] and in programming style between the OT-2 and the Prep [Linux vs. Apple]. Depending on what you need to get out of it and your desire to tinker, that may tip you one way or the other.

I’ll send you a DM so that if you’d like to dive into the weeds a bit we can look at your specific workflows/needs.

[Sabbalonn]

Hi @TanjaMelbs you in the UK or elsewhere ? We have a solution but its dependant on where you would be situated.

[CamilloMoschner]

Hi @TanjaMelbs,

Here is another conversation regarding Opentrons vs other machines:
https://forums.pylabrobot.org/t/opentron-vs-preowned-premium-liquid-handler-with-50k-budget/2662
→ many things regarding the Opentrons Flex also apply to the OT-2.

I will just add a couple of points:

Calibration
@UCantBcereus has already done a fantastic job describing the incredible pain of having to constantly calibrate the OT-2!
This was one of the best things when I switched to Hamilton machines - they are just incredibly precise and equipped with an armament of super useful sensors - the OT-2 has zero notable sensors.

Lock-In
Opentrons is… well open. You can use any labware and even change the tips you are using (though I found their own tips work best and are pretty cheap).
The Hamilton ML Prep is the complete opposite:
Hamilton provides a list of supported labware, and if you go for the Prep you lock yourself into purchasing Hamiltons super expensive conductive tips.
On top of that: Hamilton has recently upgraded their pipetting channels from compressed-O ring expansion (CO-RE) I technology to CO-RE II tech.
This only means that the end of a channel is exchanged but it has major ramifications to utility because you need to use different tips, i.e. CO-RE II tips - CO-RE I tips don’t work anymore.
If you buy a Prep now it will be shipped with CO-RE II channels, which is great.
But Hamilton hasn’t changed their tip IDs (because that way established protocols that use old IDs don’t need to be changed while the machine can still be upgraded). But this means that non-Hamilton suppliers still sell their stock of CO-RE I tips with the same ID as Hamilton sells CO-RE II tips.
→ As a result, you have to purchase the tips directly from Hamilton to ensure they are the correct ones (or go through lengthy, and potentially expensive, processes of validating 3rd party suppliers).

Liquid Level Detection
Hamilton machines, including the Prep, have amazing hardware that enables channels to detect when they touch a liquid, based on charge/capacitance and/or pressure.
For chemistry, this might be incredibly useful: imagine a viscous solution, you don’t want to immerse your full tip into it and have sticky liquid attached to the outside of your tip… dripping all over the deck of your robot on its way to its destination.
LLD allows you do immerse a minimal amount of tip, adjusts based on liquid level and enables following the liquid during aspiration/dispensation/mixing, enabling more flexible coding as well.

The Opentrons OT-2 has zero liquid level detection capabilities.

Chemistry
I am not sure about this one and it has been 3 years since I tested out a Prep, so please @Michael_Vuke correct me if I am wrong:
The Prep essentially has the same channels you find on a STAR machine. On a STAR you can activate a channel feature called “Antidroplet Control” or ADC.
This feature continuously measures the pressure inside your pipette and adjusts the plunger inside the pipette accordingly: This means that you can deal with volatile liquids - something quite common in chemistry applications.
Normally the volatility of the liquid creates pressure inside the pipette which in turn leads to the liquid dripping out of the channel. ADC automatically adjusts the pipette plunger to counteract this, maintaining the liquid in the tip.

Tbh this can be a tricky thing to get working.

I don’t know whether the Prep allows you to use this feature.

The Opentrons OT-2 has no such feature and dripping of volatile liquids is to be expected (but I have never experienced anything like it with my aqueous solutions).

Processing speed
The OT-2 is not necessarily “slow” but the fact that it can only use 1 of its 2 pipettes at a time is a major time constraint, particularly if you have a lot of different factor combinations to pipette into different wells and need to change tip every time.
If I’m not mistaken, the Prep comes with 2 y-independent channels and can be upgrade with another fixed-distance 8-channel next to the other 2.
This means you have at least 2 channels that can aspirate/dispense/mix in the same x coordinate independently of one another at the same time.
Properly programmed this can save up to 50% of your processing speed.
Also, maybe its just me, but I got the impression that the Prep moves faster in general.

Automated sample movement
The Prep has 2 grippers that can be picked up by its channels.
These allow you to move plates around without human intervention, very useful if you need to put something on the the integrated heater-shaker or on a magnet.

There is zero automated sample movement on the OT-2.

Automated deck recognition
One of the “coolest” and surprisingly useful features of the Prep is the camera that is mounted on its ceiling.
→ it recognises not just the (supported) labware on the deck but even recognises how many tips are still left in a tip_rack, avoiding you from having to manually re-rack partially-used racks which I found saves a surprising amount of time.

The OT-2 technically has a camera built into the front-left pillar… but I have never heard anyone use it or even mention it in the last couple of years.

Programmability
There is no talking around this: the Prep is, in my experience, an absolute nightmare to “programme” (if nothing has changed in the last 3 years):

1. you have to “programme” it by tapping commands into the robot-attached touchscreen. The screen could not be unmounted, leaving you to stand in front of the machine for hours until your arm hurts too much and you try again tomorrow - with a stronger arm.
2. the Prep has limited storage space. If I remember correctly, this means you cannot store more than 8 (?) automation scripts at any given time. If you want more scripts you have to delete the old ones. This goes against pretty much all good programming practices, from standardisation, abstractions, code reusability to version control.
3. every automation protocol on any machines requires testing of the steps inside it. This process is incredibly difficult when you have only a touchscreen and you have to execute all steps of your protocol that chronologically come before the one you are about to test (when sequentially building up a new protocol), and testing out steps as separate protocols isn’t really possible because of the script limitation problem mentioned in 2. above. → this means you have no “real-time control” of your script, i.e. click ‘aspirate’ and see the Prep aspirate, then click ‘dispense’ and see it dispense.
   To my knowledge this isn’t a problem for application scientist because they do have access to a scripting software that is not shared with customers.

Now compare this to the OT-2:
It has 3 modes of operation:
(1) a web-based GUI to write protocols [personally I find it horrible but I am a coder and I’ve met plenty of non-coders who found it useful],
(2) a Python/.py file execution mode, where you just drop the .py file into the main OT-2 software. LLMs like ChatGPT are already pretty good at writing this for you.
(3) you can connect directly to the Raspberry Pi that control the OT-2, open a Jupyter Notebook on it and execute Python commands line-by-line in real-time. I think the incredible user-friendliness of this mode makes it incredibly powerful for coding-literate users, and for teaching purposes.

For all 3 modes your automation scripts are just text files (either JSON, .py or .ipynb [just a JSON in disguise]) → this means you can use Git(Hub) to maintain your scripts, modify them, load old versions, distribute them in seconds and work collaboratively with other people.
Plus, you can write programmes in a café

Load files
How do you make automation throughput adaptive, i.e. process 5 samples today and 100 tomorrow with the same script?
One way: you load a spreadsheet as input.

I have no idea how the Prep would do that - though I am hesitant to say it can’t because I don’t know.

The OT-2 can easily do that through the Jupyter Notebook execution mode.

---

That all said let’s revisit your requirements:

“We need for our plans”

• high flexibility
  • this indicates OT-2 to me, programming is a breeze on the OT-2 and highly flexible
  • programming on the Prep is what stops me from buying it
• ability to fill 96 well plates in each well with individual mixtures
  • this is possible with both machines, much easier to programme with the OT-2, but potentially much faster and much more reliable on the Prep
• python control and a system that is easy to extend and integrate into other workflows as good as possible.
  • this is wonderfully specific and indicates that you have prior Python knowledge. Since you only mentioned three needs and Python is one of them there is only the OT-2.
    Note: my comments are based on the believe that you know some Python. I believe there is a lot of value for non-coders in the systems that are in place for both the Prep and the OT-2.

In summary, all hardware needs you might have point towards Prep and all software needs point towards OT-2.

but

there is a way to combine them both:

You could buy the Hamilton ML Prep, write a Python backend for it to be used with PyLabRobot, and then use Python to control it.

PyLabRobot (PLR) is the open-source Python library that gives this forum its name, and I’d call it the prodigy child of PyHamilton.
It enables Python control of not just one liquid handler but currently supports:

• Opentrons OT-2
• Hamilton STARlet
• Hamilton STAR
• Hamilton Vantage
• Tecan Freedom Evo
• a programmable scale
• heater-shakers
• temperature-control modules
• a plate reader
• a peristaltic pump
• audiofeedback from any of these machines

…and enables use of any operating system (Linux/Windows/Mac/Raspberry Pi OS).

New machines are constantly added, and personally I would probably buy a Prep if it was controllable via Python, and specifically PyLabRobot

Especially, you hinting at “easy to extend and integrate into other workflows as good as possible” means PLR might be ideal.

Plus, the PyLabRobot community is amazing, ultra-fast and, because it is an open-source project, completely free.
You can ask for a completely new feature to be made (or make it yourself) and, in my experience, could have it working in less than 5 hours.

Also, I was not aware how expensive the OT-2 has become. During my PhD, I bought one in May 2021 for $6,000 and now it is $13,500, and I bought 2x GEN2 pipettes for $900 each, which are now $2,250… I do not think the capabilities are worth this much money - which hurts me to say because I love the OT-2 as my first liquid handler.
I don’t know how much the Prep is because Hamilton prices are oddly non-transparent but I think you are right to consider it as I cannot imagine the price difference to be that much more anymore.

(Depending on your uni’s purchasing requirements, there is also always the second-hand market)

[CamilloMoschner]

Another small thing:

To make fair comparisons…

1. Keep in mind that the Prep comes pre-shipped with a Hamilton heater-shaker (to my knowledge, whether you want it or not - but it is very useful) which is by itself worth about $6k.
   So for comparison purposes, OT-2 ($13,500) + 2x pipettes ($2,250) + Opentrons heater-shaker ($3,750) = $21,750 for an OT-2 with just 2 single-channel pipettes. That is the minimal fair price to compare with the Prep.
2. The Prep has 7x SBS positions + the Hamilton heater-shaker + dedicated liquid_waste + dedicated tip/other waste.
   The OT-2 has 11x SBS positions + dedicated tip/other waste.
   But if you were to write a PyLabRobot backend for the Prep you can perform vertical stacking on the Prep because it has grippers while the OT-2 doesn’t.
   This means your deckspace can increase by about 5x with micro-titer plates stacked on top of each other times every of the 7x SBS positions (for plates), or about 2 or 3 stacks of nested tips times 7 (for tips).
   I am not aware of a physical/hardware constraint that limits the Prep from doing this but I don’t know whether the current tablet software enables any form of stacking.

[Michael_Vuke]

Thanks for putting together such an extensive response, @CamilloMoschner. I can’t speak to what the Prep was like 3+ years ago (I’ve been involved with it a little over 2 years), but I’ll share some updates to what you said to speak to what the Prep is like today.

Blockquote The Hamilton ML Prep is the complete opposite:
Hamilton provides a list of supported labware, and if you go for the Prep you lock yourself into purchasing Hamiltons super expensive conductive tips.

We do have a list of supported labware, but to clarify, this is not an exclusive list. We provide free labware definitions to get you using what you already have in the lab. We do the definitions for you so that you don’t have to worry about making an inaccurate definition (something that can be really easy to miss, and then you’re sitting there trying to figure out why your system isn’t working properly all of a sudden). This also lets you get proper liquid following functionality.

With our most recent software update, we’ve even changed some things on the backend that give us even more flexibility to define irregular/non-SLAS standard labware!

For the tip pricing point, expensive is relative, but what I will say is when I’ve done cost comparisons with potential customers, I’ve yet to have someone walk away from the Prep because of it, and people have said it was a fair price (and better than they expected).

Particularly once you weigh slightly more expensive tips vs. the cost of failed/repeat runs due to user error, lot to lot variability of other tip quality, tips falling off of a robot mid-run (if it’s a friction fit, that happens!), etc. the wasted time, samples, and reagent quickly offset the tip price

We also make our tips in-house which means we don’t have the markups some other companies have on their proprietary tips–cut out the middle man–which helps us keep our prices competitive.

Blockquote Editorial note: why do we make our own tips/make them proprietary?

If we say we’re moving 1uL, 50uL, 837uL, etc. we need to be pretty darn sure we are moving that much liquid! If we don’t control the manufacturing process, we can’t verify the tips will be consistent enough to validate those claims. We’ve seen this with other systems where a protocol runs fine until it suddenly hits errors–the lab later figures out that it was lot-to-lot variability from the OEM supplier of the tips.

By doing our CORE II tech, you get:
–Tips physically locked in, so they cannot fall off. Very important if you’re trying to walk away from your robot while it’s pipetting!
–Sealing at the exact same spot on every tip. No more of those sine wave patterns on the bottom of your multichannel. So you’re not crunching some tips while others are out of liquid. You’re also ensuring a proper seal at the proper spot, which lets you be more precise with your volumes.
–No cross contamination. If you have to forcibly eject a tip, you risk creating aerosols that can contaminate the deck/other samples. We just stop holding the tip in place and it falls off with no aerosolization.
–Tip reuse. You aren’t forcibly distending the collar of the tip, which means you can reuse those tips a lot longer than a traditional pipette tip while retaining “new tip” performance.

Okay, back to the post!

Blockquote The Prep essentially has the same channels you find on a STAR machine. On a STAR you can activate a channel feature called “Antidroplet Control” or ADC.

Great call out! The answer is yes and no.

The Prep does use the same channels as you see on the STARs (and NIMBUSs [Nimbi?]). That’s one reason the quality and accuracy is so high!

At this point in time, we don’t have ADC on the Prep–in trying to keep the software easy to use/not confusing for standard bench scientists (vs. automation engineers), we don’t have all of the button and knobs open to fiddle with. BUT if that’s something people are interested in, let your reps/me know and I can request it! I just had a call yesterday with product management discussing the future update wishlist/customer feedback.

However, with our liquid classes you can do some things to help avoid dripping.

Blockquote you have to “programme” it by tapping commands into the robot-attached touchscreen. The screen could not be unmounted, leaving you to stand in front of the machine for hours until your arm hurts too much and you try again tomorrow - with a stronger arm.

–At this point in time there are limited option for programming off of the Prep (there is the option to buy a separate “Preplet” which is the screen+internal PC). However, my personal favorite way to do programming on the Prep is to just use a USB mouse and keyboard next to the Prep. That stops the arm fatigue

Blockquote 1. the Prep has limited storage space. If I remember correctly, this means you cannot store more than 8 (?) automation scripts at any given time. If you want more scripts you have to delete the old ones. This goes against pretty much all good programming practices, from standardisation, abstractions, code reusability to version control.

I’ve yet to have someone run out of space for their protocols. On my personal Prep I have over 70 saved . On the homescreen you can have 18 favorited protocols, and then your protocol library can hold hundreds of other protocols on the unit.

Also, on your run reports it will indicate the last time that protocol was edited, so you also have some additional documentation around versions.

Blockquote every automation protocol on any machines requires testing of the steps inside it. This process is incredibly difficult when you have only a touchscreen and you have to execute all steps of your protocol that chronologically come before the one you are about to test (when sequentially building up a new protocol)

When you’re building a protocol, there are options to disable steps (but keep them in the protocol) so that you can jump straight to the part that needs to be tested.

Blockquote How do you make automation throughput adaptive, i.e. process 5 samples today and 100 tomorrow with the same script?
One way: you load a spreadsheet as input.
I have no idea how the Prep would do that - though I am hesitant to say it can’t because I don’t know.

here are two ways we can approach this on the Prep:

1. Worklists. We have several steps (hitpicking, reagent from file, and normalizations) that are explicitly built for running off of .CSV files. This lets you programmatically adjust your samples as described. Additionally, you can gain additional flexibility by combining this work listing with barcodes which means even sites on the deck don’t have to be defined, because you can use the reading of barcodes to additionally define that at runtime.

2. “How many samples?” For those who don’t like to use .CSV files, you can con control the number of samples with the “how many samples” step. This lets you create run time masks for the labware on the deck. Essentially, when you build the protocol you make a superset of functionality, but add a variable at the beginning that lets you specify how many samples are present in your tube racks or plates before runtime, and it will only run the functionality for that subset of locations on that run.

Blockquote Keep in mind that the Prep comes pre-shipped with a Hamilton heater-shaker (to my knowledge, whether you want it or not - but it is very useful)

To clarify, the HHS is sold as an optional accessory. We also have the option for the HHC (heater chiller) for active cooling applications! But it is optional so you can get what you need and not have to buy irrelevant stuff.

Blockquote The Prep has 7x SBS positions + the Hamilton heater-shaker + dedicated liquid_waste + dedicated tip/other waste.

To the previous point, without a HHS or HHC, you get 8 SBS positions. With a device you have the device + 7 positions.

Blockquote you can perform vertical stacking on the Prep.

I can’t speak to what a “Jailbroken” prep would do (other than not be covered under warranty ), but we don’t currently support stacking on the Prep, in part due to Z height clearance. We condensed the Prep as much as possible so that it could fit into BSCs/fume hoods.

---

Some other thoughts:

1. Thank you again for this run down, Camillo! I love seeing the forum in action. Hopefully the clarifications above help.

2. If anyone runs into something the Prep “can’t do”, please reach out to us! A lot of times we hear from people who tried something and moved on because they “couldn’t do it” , but after we talk with them we are able to help them execute what they hoped to do!

3. To point #2, if you can’t do it currently, what I can say is that we are actively soliciting feedback and feature requests to add to the Prep. We just dropped a big [free] software update a month or so ago that added a bunch of functionality that had been requested! Things like pipetting offsets and such. So we are always adding features and functionality.

4. The Prep is 21 CFR pt 11 compliant and gets used in regulated industries a lot! Some of the other liquid handlers in this space explicitly state that they shouldn’t be used in non-research settings. Whether or not you are in a regulated environment, that should speak to the robustness of the hardware and software.

If you can’t tell, I really like the Prep! (and I liked it before I sold it–one reason I came to Hamilton)

Let us know more questions/feedback and we’ll help however we can.

[Kadams]

There is a Prep in another lab at my company. It just collects dust and causes anger. They bought it before I was on board and have said that Hamilton has basically abandoned them and will offer no support for this device. So much so that we may ship it from the West Coast to Boston so I can work on it. Being stuck programming on the touch screen is also a real bummer or else I could help them remotely. YMMV

[EricSindelar_Hamilton]

Sorry your colleague had that experience. The Prep has a different support model and team than the larger instruments so perhaps that is part of the confusion. Users can request support through our knowledge base and ticketing system online - link is below. @Michael_Vuke is also active on this forum to help clarify functionality and to direct customer inquiries.

As Michael pointed out, you can use a USB mouse and keyboard to navigate and program instead of relying soley on the touch screen.

https://www.hamiltoncompany.com/automated-liquid-handling/platforms/microlab-prep/prep-help

[Michael_Vuke]

To expand on Eric’s comment, that’s definitely not the way we try to handle issues. The Prep has a support team dedicated exclusively to supporting it, in addition to the field sales team (recently expanded to let everyone have support closer to home).

I’ll reach out to you on the side to try to learn more about your specific case and make sure that you’re plugged in with the right team at Hamilton so that we can get you back up and running.

Like Eric said, it may be that your colleague had asked someone on the Robotics side about the Prep and was told that they didn’t support it? The Prep falls under the Lab Solutions team. Either way, Robotics and Lab Solutions work hand-in-hand to make sure that we get our customers the support they need. (I share an office with the Robotics rep, apps, service, etc. teams. @NickHealy_Hamilton and I talk constantly! So different units but we’re one org working to get y’all supported)

[Kadams]

@EricSindelar_Hamilton I think that the support model is different is a major source of the issue.
The group has been frustrated to the point that they no longer want to even pursue it. If you have a direct contact that they could reach out to that could work. Otherwise I think that ship has sailed.

[luisvillaautomata]

I know a company that has “hacked” the ML Prep so that you can control it with Python. It gives you more flexibility and they’ve developed an experimentation platform around it.

If you’re interested in connecting with them, send me a message.

[CamilloMoschner]

@Michael_Vuke

Thanks for putting together such an extensive response, @CamilloMoschner

You are very welcome, @Michael_Vuke, and thank you for your clarifications.
It is clear the Prep has evolved a lot. It sounds like the internal computer has been upgraded significantly.
I didn’t think about using a keyboard and mouse. To be honest, I wasn’t even aware Instinct would support them.
Does Hamilton offer a guide to ergonomic working when using the Prep?

@Michael_Vuke

Also, on your run reports it will indicate the last time that protocol was edited, so you also have some additional documentation around versions.

Can you recover an old version of an automation script?
And, is there Ctrl+Z/“undo” support or is a change to a script permanent, with no opportunity to go back a step?
Also, how can users transfer automation protocols from one Prep to another Prep (e.g. in the same company)?

@Michael_Vuke

Worklists.

Can you please elaborate on the mode of file transport: How is a worklist spreadsheet moved onto the Prep?
Is the Prep now connected to the internet and a cloud service for file transport?

[CamilloMoschner]

@CamilloMoschner

So for comparison purposes, OT-2 ($13,500) + 2x pipettes ($2,250) + Opentrons heater-shaker ($3,750) = $21,750 for an OT-2 with just 2 single-channel pipettes. That is the minimal fair price to compare with the Prep.

@TanjaMelbs I looked into these numbers regarding Opentrons pricing again because they just didn’t seem right to me:
the bare-bone OT-2 price going from $6,000 (May 2021) to $13,450 (March 2024) just didn’t sound like the Opentrons I know, and I was indeed wrong:

The OT-2 at $13,450 already comes with two single-channel pipettes.
The price without pipettes is $8,950 which, to be fair, is still an almost 50% price increase in just 3 years and far above US inflation (i.e. average inflation, not any specific sector’s inflation - electronics might be higher) but product prices are the result of an amalgamation of factors, not just inflation, including a potential aim to not compete with newer products, and $9k for the OT-2 is something I personally consider to be okay… ish for the empowerment to programme a liquid handler in Python by manufacturer design.

Plus, now that we know the Prep does not have to be purchased with a shaker, the OT-2 shouldn’t be compared with purchasing one either.

So, a fully operational OT-2 with 2x pipettes is $13,450 for comparison purposes with the Prep.

---

You should also be aware though that the Prep, because it uses Hamilton STAR channels, can pipette a theoretical volume range from 0.5-1,000ul, using the same channels and only changes to specific tips for certain ranges (to my knowledge: TIP_50ul: 0.5-50ul, ST/300u; 50-300ul, HT/1000ul: 100-1000ul).

The OT-2 requires special pipettes for special volume ranges + the pipette-specific tips:

And since you can only install 2x pipettes at a time you have to choose the permutations of possible pipetting ranges you want to use for a given automation run. This does increase the planning requirements for the OT-2 and makes it less flexible than the Prep in this regard.

---

Hope these bits of information help you make an informed decision about the tradeoffs of each system.

[Kadams]

One really good reason to get an Opentrons vs Hamilton Prep is you can get Genie LabOS. Its really great software and makes everything easier.

[Michael_Vuke]

Does Hamilton offer a guide to ergonomic working when using the Prep?

Can you shoot me a DM with the sort of information that would be useful for that guide? We are continually building out even more support resources/guides, so user feedback is always welcome. Don’t want to clog up this thread with that though, since it’s off topic for OP’s question.

Can you recover an old version of an automation script?

The Prep reminds you to backup your system at given intervals (and you can backup more frequently), so you should have a consistent copy of all your protocols as they were at various dates.

We don’t currently have an “undo” button.

how can users transfer automation protocols from one Prep to another Prep (e.g. in the same company)?

Can you please elaborate on the mode of file transport: How is a worklist spreadsheet moved onto the Prep? Is the Prep now connected to the internet and a cloud service for file transport?

You can use the built-in export function to share between Preps via USB or local network.

Similarly, you can import files with the USB or local network. (if same file every run, you upload at time of programming. If a different file every run [most common], it prompts you at run time.)

It is not connected to Wi-fi/Cloud, in part due to user feedback. Many companies/agencies either are not allowed to have devices on the internet or have to jump through many additional hoops on instruments that do, so at this point we have kept it to local networks and USB.

---

I would definitely recommend that anyone interested in the Prep reach out to their local Hamilton team (if you don’t know who that is, DM me, Eric, Nick, or Brandon and we can connect you) to discuss its features and capabilities.

I’ve found from talking with a lot of labs that there is a perception of what the Prep can/cannot do or what features it has/doesn’t have that often doesn’t match the Prep’s capabilities. This is completely understandable! There are new features being added and things being refined constantly, or things that have been there but maybe not fully explored by a given team. Y’all’s job isn’t to be the expert on every facet of every system out there–that’s an impossible task! That’s the job of me and other reps for the companies.

So please talk to other labs and learn what their experiences are/were, but also talk to the companies directly to confirm if any capabilities or limitations you’ve heard of exist, as we (and I’m sure most companies) do constantly evolve

[ben]

if the ML Prep had pylabrobot support, I think it would easily be the best benchtop liquid handler on the market