Perfect Pipetting – Part 2

 The microlitre mystique of accurate volume transfer

Introduction

Achieving precise and accurate liquid volume transfers is the cornerstone of reliable experimental results. But it’s not often that said experiments use water and only water with the same transfer workflow. If you want to know about water transfers, be my guest and follow the link. But I think this is far more interesting. Here is a lot of data from multiple similar experiments. 

Method

Manual Gilson pipettes, piloted by an experienced user were compared to SQWERTY, piloted by a user in-training. Said users were the same person. All pipettes were calibrated and used with compatible  200 μL tips to transfer 180 μL of liquid (unless otherwise stated) into a 96 multi-well plate (MWP). This was done in triplicate. The weight change was calculated to estimate the average volume per well over the entire 96 MWP. This was presented as the percentage of total error averaged across the replicates. A positive error means that too much liquid was transferred, and a negative was too little.

Multi-dispensing water

The SQWERTY multi-dispense operation aspirates a large volume of liquid from one source and dispenses equal smaller volumes into target wells. This exact step can’t be replicated with manual pipettes, but can be compared to an extent. Figure 1 shows the percentage of total error when 40 μL water was transferred by manual pipettes compared to SQWERTY using the multi-dispense operation. 

Figure 1: The average percentage pipetting error when 40 µL water was transferred into a 96 MWP by a Gilson P200 manual pipette, P200 8-channel pipette and the SQWERTY multi-dispense workflow (default settings). All three pipettes yielded good results, but it was SQWERTY that produced the lowest error (0.02%) and with very little variation. There was an estimated 6 μL difference between the highest volume per well and the lowest across all the replicates.

Transferring YPD broth

Of course, a lot of researchers may want to transfer nutrient broth into a MWP. For this experiment, YPD broth (for yeast culture) was transferred by SQWERTY using the pipetting settings and workflow designed for water, i.e. the default settings. The kind of workflow one might quickly program when filling a MWP prior to inoculation. 

Figure 2: The average percentage in pipetting error for YPD broth volume dispensed by a Gilson P200 pipette and the pipetting robot SQWERTY (default pipetting parameters). The manual pipette shows four times more error than SQWERTY with a greater degree of variation in volume delivered to each well.

Transferring the more difficult to pipette solutions

Anyone who’s ever worked in a lab knows that THE most difficult solutions to pipette are highly concentrated glycerol (reeeeaaally sticky and messy) and ethanol (like a toddler – can’t keep it in one spot). So to put SQWERTY to the test and push the limits, that’s what this experiment transferred, again using the default settings designed for water so a clear starting point could be established.

Figure 3: The average % error in volume transferred into a 96 MWP of 75% ethanol (A) and 75% glycerol (B) by a Gilson p200 pipette compared to the default SQWERTY transfer operation. It is clear that in All cases the volume accuracy is hugely better when using SQWERTY compared to the manual pipette, and the consistency across the replicates is comparable (glycerol) or better (ethanol).

When manually transferring glycerol or ethanol, most lab rats would alter their usual pipetting technique (we did, too). Highly concentrated glycerol and ethanol are more viscous and volatile, respectively, than water, which makes a big difference in volume accuracy. Fortunately, SQWERTY also has a feature to optimise pipetting parameters for more or less viscous liquids: Interactive mode! To see what happened after we optimised glycerol pipetting, check out this article. From Figure 3A, it looks like ethanol might not need much further optimisation, with an average error of 0.6% but you might have a different opinion based on your application. 

Conclusion

The first thing that you learn about volume accuracy in automated liquid handling is that it’s affected by everything … EVERYTHING!!! Some things can’t easily be controlled, like the atmospheric pressure, humidity or the temperature in your lab. But others can. This data shows example differences for different liquids and workflows. 

A minor difference in liquid volume can cause approximately 10% imprecision to experimental results (Lippi et al., 2017). This is why optimising and validating your workflows is be important. As a starting point, using the default settings, this data shows that the pipetting robot SQWERTY is already better than using a calibrated manual pipette.