Biomek 2000

Posted by Reto Hoehener on September 24, 2012 | Laboratory Automation

For the impatient: The following short videos show the Biomek 2000 in action.

 

Background

In February 2012, after having read Aubrey de Grey's book Ending Aging, I immediately started to wonder how I could contribute to the SENS cause. Having a computer science background, I was unsure of what this kind of research consisted of. I quickly discovered that the SENS Foundation ran it's own research center and decided to apply as a volunteer, in order to get a better feeling for this exciting and challenging field. I count myself very lucky that, even though I have no background in biochemistry, David Halvorsen agreed to take me on as a volunteer. One of the projects he had in mind for me was to get this liquid handler up and running.

 

Acquisition

In April 2012, David was able to find a very reasonably priced used Biomek 2000 on GoIndustry DoveBid. The lot sold for just $200 and was delivered to SENS shortly afterwards.

The auction:

 

Connection problems

The robot seemed to be in good condition: It powered up, fans turning, leds flashing. But, it was not possible to get the software to talk to the Biomek (and vice versa). Various causes were suspected: Faulty circuitry inside of the liquid handler, defective serial ports on the machine as well as on the computers used to connect to it, a wrong or defective connection cable, incompatible software versions, incompatible operating systems etc. When I arrived at the end of July, I could do nothing more than repeat the previous findings: Inability to communicate with the Biomek. I had tested with three different serial cables, systematically going through all possible port and baud rate combinations, even asked for help on laboratory automation mailing lists and forums – to no avail.

Only a stroke of luck almost two months later changed this situation. A representative of Beckman Coulter (the manufacturer) came by the research center one day and noticed our Biomek sitting in the lab. She asked a few questions about the robot and offered to ask her engineers for advice. Not only did she do that, but even offered to come by with a support engineer a few days later to have a quick look at the Biomek. We were reluctant to take them up on this offer since we had no budget for this project, but they offered to come by anyway, for free! Only three days later, a Beckman Coulter engineer was looking at our Biomek and immediately asked what kind of cable we were using to connect to it.

Turns out we were using the wrong cable. The documentation only mentioned a "connection cable", but the ports seemed to indicate a plain serial cable (RS232). Wrong assumption: What was needed was a so called null-modem serial cable. Armed with this critical information, Ann picked up a null-modem serial cable for $5 at a local electronics shop the next day and, oh wonder, I was immediately able to connect to the Biomek - happy moment.

Null-modem serial cable:

 

Hardware and Software Setup

For those interested, this is the current working setup:

  • The Biomek 2000 Workstation
  • PC running Win XP, with serial port
  • BioWorks Version 3.1a (the Biomek controlling software)
  • The infamous null-modem serial cable to connect the Biomek to the PC (using Port 1 on the Biomek)

Both the hardware and software reference manuals are available for download from the Beckman Coulter website.

The Biomek 2000, with the Win XP machine to the left:

 

Alignment

The manual requested that the Biomek be "aligned" weekly and especially after relocation. This is a short procedure where the alignment tool and a pipette tip is attached to the robot arm and guided to several calibration points on the work surface. The goal is to teach the Biomek the exact positions of the labware holders.

The robot arm with the alignment probe (red). The pipette tip is pointing at a calibration point on the labware holder:

 

PCR reaction plate preparation

Disclaimer: I am by no means a PCR specialist. Every PCR experiment is probably set up in a unique way. The following just represents one possible way, as explained to me by Ali Crampton (thanks!).

One task where a liquid handler can come in very handy is the preparation of PCR reaction plates. The manual pipetting involved can easily take up an hour or two.

At the time of writing the lab was trying out a new real-time PCR machine called StepOne (Applied Biosystems). It uses a 48-well reaction plate, so I decided to try and setup a Biomek method to prepare this specific plate.

The StepOne PCR reaction plate after a test run. Each well contains 20 uL of water. As part of the test every other row was mixed with red food color. The streaking was a result of the food color being extremely gluey:

The input for the reaction plate is prepared in 0.5 mL microcentrifuge tubes with caps. One set of tubes contains the master-mix (PCR reagents, primer, water), the other set contains the DNA. An easy layout to keep things organized is to distribute every type of master-mix throughout a column and every DNA throughout a row. In this case, I assumed each master-mix tube to contain enough liquid to dispense 18 uL into all 6 wells of one column, and each DNA tube to contain enough liquid to distribute 2 uL into all 8 wells of one row of the 6x8 reaction plate.

Master-mix / DNA distribution plan:

 

Biomek labware setup

The only original labware that came with the liquid handler were the P20 tip boxes. I created two improvised tube racks by placing empty tip holders from used tip boxes on top of standard 96-well plates. The 96-well plates we had on stock in the lab luckily fit pretty nicely into the labware holders (they have a wiggle room of about 1mm). The advantage of using 96-well plates as a basis is that the Biomek software already has many existing 96-well plate definitions on file. The software allows to derive new labware definitions by copying an existing one. In this setup, the robot thinks that the tube racks are 96-well plates. The well center positions are identical - the only values I had to adjust in the custom labware editor were "well depth" and top height.

The labware editor showing the custom tube rack definition:

The worksurface setup as shown within the BioWorks software:

The improvised worksurface setup:

The 500 uL microcentrifuge tubes are placed in a zig-zag pattern because they didn't quite fit next to each other in a single row. The BioWorks software supports these kind of setups with "custom patterns".

A custom pattern:

 

Method configuration

The pipetting instructions for the Biomek are defined in the BioWorks Method editor. In terms of BioWorks language, this process only needs two pipetting steps. The first pipetting step transfers liquid from the top tube rack, the second one from the left.

The method steps:

The interface gives the experiment designer complete control over every aspect of the pipetting: Volume, aspirate and dispense height/speed, when to change tips, mixing etc.

The first pipetting instruction:

 

Conclusion

For me personally, the main conclusion is that it is only worth automating a process like this if it will be repeated in an identical configuration many times over a period of at least half a year. As soon as the individual experiments vary slightly, the automation effort will be bigger than just doing it manually.

I was positively surprised by the pretty straightforward way that the software allows you to setup your experiment in any way you wish.

Regarding precision: I found that the P20 pipette tool is able to consistenly dispense volumes of 10 uL to the bottom of the reaction wells. Although, it was necessary to fiddle with the amount in the software, e.g. configure 11 uL in the software to get 10 uL in the well. It was also necessary to dispense very close to the bottom of the reaction well so that the liquid would stick there (and not to the pipette tip).

Trying to pipette very small volumes like 2 uL was not very successful. When dispensing from the air, the droplet tended to just hang on the pipette tip. Dispensing while immersing the tip into an existing mixture resulted in anything between 0 and 1.5 uL being added to the mixture.