Fourth Annual DNA Grantees' Workshop

Tuesday, June 24, 2003

MORNING SESSION

High-Throughput Testing and Analysis
David Coffman
Biography

MR. COFFMAN: My topic is on high-throughput nature. As you know, in Florida we've been using automation in some form since 1996. We went out to California Department of Justice and basically just mimicked what they were doing, because we were in emergency mode. Since that time, we have developed different techniques and different high-throughput procedures as we needed.

The reason we didn't call this databanking is because caseworking labs, at least in Florida, are finally accepting the fact that casework can be automated. You may not be able to automate it all at one time, but eventually you're going to be able to automate a lot of your casework samples. We're doing it in Florida. As you heard, Louisiana has already performed it. So I think it's crucial to look at that.

My goal is to tell you what we have done when shopping for automation and high-throughput technology. Sozer: Slide 7

Backward compatibility is always important to a researcher, because most labs are not going to want to jump on new technology unless it's somehow backwardly compatible with the technology they already have. As you know, last spring we hit 1 million platforms in the national database for offenders and a lot of casework. It's going to take a huge benefit to reanalyze everything before we just change the database again, I think.

Backwards compatibility is important, but it's not the only thing. There are benefits for Y-short tandem repeats and research and that's very, very crucial.

Cost is a big issue. I don't just mean the dollars and cents per sample. You also have to think about reagent costs, which adds to your cost per sample, and instruments and the scaleability of the instruments. Also, all these liquid-handling instruments are really good, but you have to look at what your State potential is. I say "potential" because 5 years ago when we were getting a lot of samples and Virginia was the first to go to all felons, a lot of States thought that they would never have more than 1,000 a year. Now those States are getting 20,000 and 30,000 a year. So you need to look to your future potential, because these machines last quite a while.

You don't want to necessarily get a system that takes one robot to do two 96-well plates. You want a system and the chemistries that are at least the kind that you can stagger and get more samples or more plates onto a robot. For instance, the robot we're currently using, we needed eight 96-well plates a day just to keep up with what was coming in the door. Recently, however, we staggered our procedure and are now doing 12 plates a day on just one robot. I think four or greater plates is what you need to really look at, because you don't want a lot of robots. You don't want to keep buying robots because the costs of service contracts for each robot add up. There's always something going down. So that's something you want to consider.

I think you really want comparable or better results than the manual process. By that I mean you want to be as close as possible to what you can do manually before you switch technology, especially as you move toward working with casework samples and that sort of thing.

Contamination is another crucial issue. We've done contamination studies on each of our robots, as well as a whole washing cycle. How long do you wash it between each probe if you're not using a disposable tip? You want to watch that contamination issue. Obviously, accuracy and sensitivity are key: You don't want to start losing sensitivity, especially with casework samples.

You'll also want to consider the ease of use and reliability of the instrument components; this is more toward the researcher side. As you are working with your research and wanting to partner with companies to help fund your research, make sure these companies have a good track record with customer service. Nothing's worse than having a system that's been linked to a particular platform and then you get it and you find out that it takes customer service 2 months to do preventive maintenance or repair on the robotic platform or some component of that platform. Be cautious and check the customer service track record for instrument components and pay attention to what other customers have said about how well the manufacturer responds to problems. We've had as many as 6,000 samples a month. With your equipment down at that rate, it takes you months to recover. You just want to make sure that you can get your instruments back up when they're down.

Thanks a lot.

DR. SOZER: Thanks, Dave.