Automated Colony Pickers Evolve

Everyone knows that the first genome sequencing projects took years of work and represent the combined product of tens of thousands of individual fragments.

Jul 4, 2005
Helen Dell(

Courtesy of Richard Summers

Everyone knows that the first genome sequencing projects took years of work and represent the combined product of tens of thousands of individual fragments. But what people may not have considered is that before any of those fragments could be sequenced, they first had to be cloned, picked from a library, and mini-prepped. Picking all of those colonies by hand would have required an army of technicians so, in the early 1990s, the sequencing institutes decided to automate.

"We asked our associates what were some of the onerous tasks we could help them to automate, and this was right at the top of them," recalls Joe Jaklevic, who headed up the team that designed one of the first colony pickers at the Lawrence Berkeley National Laboratory in California, where much of the human genome was unraveled.

Since then, automated colony pickers have morphed from large, single-purpose, custom-built machines to relatively compact, off-the-shelf units designed for a whole range of applications. They're expensive, to be sure, but not prohibitively so. The good news is they can work all night and never get repetitive stress disorder. Just think of them as the unflinching, eagle-eyed robotic assistants you never had.


Colony pickers image culture plates using illumination from beneath the plate. An attached computer then processes that image to locate the colonies and instructs a robot arm tipped with a pin to stab those colonies and transfer the bacteria to growth medium in a particular chamber in a multiwell plate. The pin is then sterilized in ethanol or peroxide and with heat, and the cycle begins anew.

The concept hasn't really changed much since the first design, according to Jaklevic. "There's been a certain amount of evolution in the design," he says. "But given the problem there are only so many ways to attack it." The big innovation was the change from single pins to "picking heads" bearing many pins, each fired by separate controls, so that the time the robot arm spends moving between the agar plate and the wells with the growth medium is reduced.

The Wellcome Trust Sanger Institute, Cambridge, UK, where huge tracts of the human genome were sequenced, still runs its five colony-picking machines pretty much full time. "Together the five machines pick about 60,000 colonies per day, in an 8-hour day," says Don Powell, the Sanger's press officer.

The robots each have a camera, lamp, a picking head bristling with 48 pneumatic pins, and a detection system that recognizes the color, size, and shape of the colonies. If colonies are too close together, the system disregards them in the interest of accuracy. "The resolution in the x-y plane is less than 0.01 mm," says Powell.

Several commercial colony pickers work at similar rates. The QBot from Genetix of Hampshire, UK, for example, has a 96-pin head that picks an advertised 3,500 clones per hour, with greater than 98% yield. The reality is not quite that good though, according to Andrew Paterson, director of the Plant Genome Mapping Laboratory at the University of Georgia in Athens. "Calibration between plates takes some time," he says. "We perhaps actually realize about 2,000 clones per hour." This is still a massive throughput, and he's pleased with the machine – his second from Genetix. KBiosystems' K3 Genomic Platform is even faster, according to the manufacturer. With two 96-pin heads – one picks while the other is being sterilized – the K3 can pick 4,000 clones per hour.

Slightly slower in the speed stakes are a range of smaller machines for those with modest aims and less space. The VersArray Colony Picker and Arrayer from Bio-Rad Laboratories of Hercules, Calif., for example, can pick 850 colonies per hour, and measures 1.43 m H × 1.27 m W × 1.02 m D (the Genetix QBot measures 1.96 m × 2.02 m × 1.58 m). And the Genomic Solutions' BioPick is just 0.9 m × 1.1 m × 0.4 m, and can pick around 1,000 colonies per hour. [In comparison, the Sanger's platform is 2 m × 2.1 m × 1.3 m.]

Though the price tag for these machines is fairly steep (the list price for the VersArray Colony Picker and Arrayer, for example, is $91,800 US), several lab groups can band together to make them more affordable, as several yeast groups at SUNY Upstate Medical University did to buy a VersArray system. "We have about six yeast groups in our building, all of whom use it," says Bob West of the department of biochemistry and molecular biology. "It makes a lot of sense to have it. Nobody would want to do what this robot does day and night."

The main costs with running the machine are plasticware and consumables, according to West's colleague, David C. Amberg. "OmniTrays have been made artificially expensive due to patents, and the fact that our machine will only take Nunc OmniTrays has made the costs burdensome," he says. But overall, given that the machine is busy about 50% of the time, he reckons the investment has been worth it.

Bio-Rad says the newer-model VersArray Colony Picker and Arrayers can be programmed to deal with a range of plasticware to overcome this problem. "The field service engineer who sets up the machine will calibrate it to the plates you want to use, and then it's fixed," says Rhonda Henshall-Powell of Bio-Rad's technical service team.


Despite the name, colony-picking robots don't just pick colonies. Sometimes it's necessary to replicate a plate, rearray clones from several plates into a single plate, or "cherry-pick" clones for maintenance and viability assessment. In such cases the robots can dip their tips into stock cultures, rather than colonies on plates, before inoculating fresh cultures.

Selected Suppliers

B-Bridge International

Bio-Rad Laboratories


Genomic Solutions



Many new machines (Genetix's QPix and KBiosystems' K3, for instance) come bundled with data-tracking software and barcode readers, so tagging plates with barcodes allows users to ensure each culture's provenance. One can then instruct the robot to cherry-pick cultures A3 and B6 from plate 1; C8 and D5 from plate 2; and so on, and the robot takes it from there.

The machines often work even faster in this mode. Rather than picking 850 colonies per hour, the VersArray "pins the complete set of approximately 4,800 yeast deletion strains to a new set of plates in about one and a half hours," West says.

And the colonies of bacteria or yeast don't have to be transferred to plates, either. They can be spotted onto a filter for a hybridization experiment (called macroarraying), or to a glass slide to create a microarray, creating a large number of potential applications for these machines. Indeed the varied applications have become so integral to the machines that several companies make a basic robot platform that supports them all, and mix and match components to suit the customer's needs. Bio-Rad's Vers-Array Colony Picker and Arrayer, for instance, has both picking and rearraying capabilities, while the VersArray ChipWriter Pro microarrayer can also pick colonies.

"Some or all of the applications can be supported on our basic QPix platform," says Ian Taylor, Genetix's business unit manager for biopharmaceutical technologies. "Based upon that, we've done different sort of flavors of the product for different market segments." Take, for example, the QPExpression, which couples three related microbiological techniques – plating a transformation, spreading the culture, and picking colonies – in a single instrument. Genetix's QPix2XT, on the other hand, blends colony picking, rearraying, and gridding applications for directed evolution applications.

Some instruments use fluorescent light to identify desired colonies (if the colonies express a fluorescent protein, for instance). "This allows additional intelligence into the picking process so you can do more of the decision-making before you go in," says Taylor. "You can see on the culture plate which are the most productive and potentially valuable clones." Genetix's GloPix, and KBiosystems' K3 and K6 systems all have this capability.


If you think you need a colony picker, but don't think you can afford one, it might be time to take another look. Companies are designing these machines around academic labs and start-up companies with smaller budgets in mind.

"Most colony pickers out there are between €50,000 and €100,000 [about $90,000 to $180,000 US]," says Sean Scotcher, future developments manager at KBiosystems, based in Essex, UK. "So we are currently launching the [K6 BiOcto-Pik] system around €28,000 for people that haven't quite got the budget in place for the larger capital purchases."

But that lower price comes at a price, as it were. The K6 BiOcto-Pik picks just 1,200 colonies an hour, less than one-third of the K3's speed, largely because of its small, 16-needle picking head. But the machine will pick from all standard plate sizes, Petri dishes, and bioassay trays, which should keep consumable costs down.

Like its larger counterparts, the unit can pick bacterial and yeast colonies and phage plaques, select colonies on the basis of color and fluorescence, and has user-definable parameters for colony selection and needle sterilization. And it supports such applications as basic picking, rearraying, and replication; microarraying; and coring plugs from two-dimensional electrophoretic gels.


Though originally designed to deal with bacterial and yeast colonies and phage plaques, colony pickers are now being built to handle the demands of mammalian cultures. "Mammalian cells require gentle handing, so you need slightly different mechanics and picking tips," says Genetix's Taylor. Also the levels of sterility and containment have to be of a higher standard.

Genetix's new ClonePix instrument can be fitted with a fluorescence mode to ease picking of colonies producing fluorescent proteins. Taylor says he hopes the machine will compete with fluorescence-activated cell sorting as a screening mechanism. "ClonePix is flavor of the month with all the biopharmaceutical company customers," he asserts. "A lot of pharma companies are realizing the small molecule drugs are not necessarily the answer, and they are getting directly involved into therapeutic proteins and antibodies."

It seems that as they started, so automated colony-pickers are developing – with biology firmly at the forefront of their design, and with the technology enabling massive leaps in experimental possibilities.