How To Automate Your Lab, Part 2: Workflow And Precision

One of the biggest factors to consider in lab automation is workflow - liquid volume transfer, number of steps, and precision. Here's how to figure it out.

This post is an excerpt from our 18-page "Introducing Automation To Your Lab" e-book. Click here to download the e-book.

Chapter 2: Choosing The Best Workflow To Automate

To really dig in, let’s consider what makes a good workflow for automation—and what doesn’t. The key general concepts can be applied to almost any protocol. To get started, review “Figure 3: Your Workflow is a Good Candidate for Automation if...” table below and compare that to the workflow you’re considering for liquid-handling automation.


If all five features in the table are true for a common workflow in your lab, you should automate it! If some of them are not true for the workflow you are considering, it is probably most practical to stick to manual pipetting.

there are many good work-arounds that can make tricky manual steps automation friendly

Let’s think about these features in a bit more detail. First, you should aim for an automation solution that will easily accommodate the liquid volumes you use most often in your workflows—if this is between 1μl and 1000μl, you’ll find a lot of automation options available for you. With smaller or larger volumes, there are robots that will work, but they are usually more expensive or may only work with more restricted use cases.

Second, you should apply automated liquid handling to processes that are troublesome, time-consuming, error-prone, and monotonous.

Third, the automated workflow should be one that a lab runs often enough—at least once a week—to really justify the transition to automation. Less than this and your return on investment in both dollars and time will take awhile to become net positive. If you do it once a week, automating a workflow can pay back the initial investment in both time and money in 8 - 12 months. If you automate a workflow daily, some lab robots pay for themselves in a matter of weeks.

Fourth, you need to consider how much of a given workflow can be automated within your budget—how long you can “walk away” from the labwork. You can buy robots with automated incubators and centrifuges, for example, but these are too expensive for most labs. However, it is important to keep in mind that just because you use a centrifuge 5x in a manual protocol to spin things down doesn’t necessarily mean you need to run it the exact same way on a robot; there are many good work-arounds that can make tricky manual steps automation friendly (See “Figure 4: Translating Manual Processes to Automation”).


Lastly, 5% precision with respect to pipetting volume needs to be enough for the workflow in mind. 5% precision works best for low volumes, but we’ve expanded on that in the “Precision and Lab Automation” chapter.

A final point about workflows: sometimes what you’re trying to do is actually totally unique to your lab. Maybe you want to integrate a custom piece of data analysis software that can generate new automated runs based on data coming from your robot. Perhaps you are developing a new hardware module that you need to work on the deck of a robot. Or, you just want to be able to use the core mechanisms of a robot in a way they weren’t exactly designed to do. In all these cases, you need a custom automation solution—and there are two ways to make one: pay someone lots of money to do it, or do it yourself (DIY). DIY lab automation is on the rise, according to this 2019 Nature article, and open-source platforms like Opentrons make completing a useful custom automation solution yourself orders of magnitude easier than it was before. You can get a totally custom automation station, but it will take more resources to get running than an off-the-shelf solution for common wet lab protocols.

Chapter 3: Precision And Lab Automation

To explore the 5% precision constraint a little deeper, let’s look at a specific pipetting volume and platform. Assuming an accurate pipetting volume of 1 microliter (μl), for example, the Opentrons OT-2 pipetting robot has a 5% coefficient of variation. That means that for a programmed volume of 1 μl, the actual pipetted volume of each transfer will be between 0.95–1.05 μl—somewhere within 5% of 1μl. On most robots, the coefficient of variation improves for larger volumes.

This is, again, a place where you can pay for performance. In general, automation solutions that can do better than 5% precision at low volumes are well out of most labs’ budget, with capacitive sensing robots like the Tecan Freedom EVO and Hamilton Microlab STAR series starting around $75k-$100k, and acoustic dispensing robots starting near a quarter million dollars. Less expensive air-displacement pipetting mechanisms generally stick to this 5% precision point—including manual pipettes.

It’s worth noting that many air-displacement manual pipettes quote a better CV at 1μl, with some claiming values as low as 2.5%; however, these results are a best-case scenario and can be difficult to achieve depending on the personnel and techniques used. Additionally, a full set of top-of-the-line manual pipettes can be as expensive as some robots. When you factor in human error and inconsistent pipetting technique across lab members, an air-displacement pipetting robot will still prove to be the most precise pipetting solution available for most jobs.

Much of the information about precision presented in this guide is focused on volumes of 1μl because this is the lowest transfer volume needed in many molecular biology workflows. While you can easily automate transfers at volumes above 1μl, it starts getting complicated—and expensive—at lower volumes. Surface tension and capillary action make liquids behave differently at these volumes, and they require special handling by different hardware. For example, microfluidic devices are super efficient for specific workflows where you can use a bespoke fluidic chip. While more flexible digital microfluidics technologies are just starting to become available to scientists due to companies like Volta Labs, microfluidics is still a rigid and inflexible solution for most applications. More specialized and expensive robots can move sub-microliter volumes using acoustic dispensing, like the Labcyte Echo, or by using positive displacement, like the TTP Labtech Mosquito—but these solutions are typically only workable for labs with a lot of resources and/or very niche workflows.

This post is an excerpt from our 18-page "Introducing Automation To Your Lab" e-book. Click here to download the e-book.