While we all may agree that our liquid handlers must be verified—bad data has become increasingly expensive in drug discovery, and may cost upwards of $100,000 per day1 if carried through to the pharmaceutical manufacturing process—what is the best approach? For labs dispensing multiple solution types (e.g. aqueous, dimethyl sulfoxide (DMSO), or serum), there’s a temptation to save time and money by calibrating different liquid classes with a single solution type (aqueous, in most cases). Unfortunately, this is not an ideal approach, and could result in significant differences between target and actual dispensed volumes, leading to bad data and, thus, costly errors during downstream development processes.
As a common organic solvent, DMSO has physical properties that are quite different from water—density, viscosity, and surface tension, to name a few— which leads to the two liquids behaving very differently from each other when handled by both air displacement pipettors and acoustic-based liquid handlers. While most automated liquid handlers (ALH) come supplied with predefined liquid classes to help optimize transfer of a specific liquid type, this does not mean that the final dispensed volume will be the same across all liquid types.
Here are the results from a study2 we did that illustrate this point. For this study the default water liquid class was used to transfer all aqueous solutions. For DMSO solutions, the default DMSO liquid class was used. A “universal” pipetting technique was utilized for both liquid classes, where only the calibration factors were adjusted for optimization. We first optimized an ALH for DMSO solutions using the default DMSO liquid class, and then used those calibration factors to pipette both DMSO and aqueous solutions. Then we did the reverse—optimized the ALH for aqueous solutions using the default water liquid class and then used those calibration factors to pipette both water and DMSO. Finally, we checked the accuracy of the transfers using the MVS Multichannel Verification System.
Figure 1. ALH optimized for DMSO dispense. The resulting calibration factors were used to test both aqueous and DMSO solutions (using the appropriate liquid class) over a target range of 0.25 to 10 µL on the MVS.
Figure 2. ALH optimized for aqueous dispense. The resulting calibration factors were used to test both aqueous and DMSO solutions (using the appropriate liquid class) over a target range of 0.25 to 10 µL on the MVS.
Figure 3. Percent difference between aqueous and DMSO dispenses as measured using the MVS, after optimization with aqueous and DMSO solutions.
You can see from Figure 3 that the difference between aqueous and DMSO dispenses on an ALH (which was optimized for use with DMSO) ranges from about 5% at a volume of 10 µL to over 168% at the lowest volume of 0.25 µL. Though the appropriate liquid class was used for each solution, when the dispense routine was optimized using a different solution type the liquid handler performed quite differently. For users dispensing ultra-low volumes when screening drug candidates, such a level of error (potential inaccuracy of >100%) could be detrimental to their results.
I regularly work with customers to help them develop verification methods that best mimic their liquid handling routines. This includes dispensing with a solution type similar to what they are working with, as well as testing over the same volume range. Results can easily be affected by factors as seemingly trivial as tip type, plate type, and liquid class. When working with customers in their labs, we recommend that they verify their liquid handlers using a liquid type similar to what they normally dispense (whenever possible) to get results that reflect the true performance of their liquid handler.
I recently conducted a service for one of our customers, and before we began testing their liquid handler performance, we worked with them to determine the following:
Basically, I want to know if, for example, they transfer 100 nL DMSO with one tip type and 5 µL aqueous with a different tip type. If they do, I want to test with DMSO solutions over the lower volume range and aqueous sample solutions over the higher range using the same tip type that the customer would normally use.
The study we did certainly confirms that it is a best practice to calibrate liquid handlers using the specific liquid class normally dispensed whenever possible. While calibration with a single liquid type will help to determine overall performance of a liquid handler, calibrating with each solution type that is used (as well as tip type, plate type, etc.) should provide the most reliable results.
As a Product Development Scientist on the R&D team at Artel, Rachel Parshley has managed important product enhancement projects, including the development of 100% DMSO solutions for use with the MVS. Rachel is also part of a technical team delivering Artel Liquid Handling Service in the real conditions of customers’ labs.
(1) Hentz, N. 2016 Feb 8. The Impact of Bad Data. Artel Digest [accessed 2017 Mar 27]. https://www.artel.co/impact-bad-data/.
(2) Parshley, R.; Bradshaw, J.T.; Albert, K. Pipetting Comparison of DMSO Versus Aqueous Solutions as Measured Using Dual-Dye Technology. Presentation given at SLAS 2014, San Diego, CA; https://www.artel.co/resource-library/pipetting-comparison-dmso-versus-aqueous-solutions-measured-using-dual-dye-technology/.