Fourth Annual DNA Grantees' Workshop
Tuesday, June 24, 2003
MORNING SESSION
Application of a Single Robotics Platform to the Combined Purification, Quantitation, and Amplification of DNA in Forensic Samples
Pat W. Wojtkiewicz
Biography
MR. TAMBASCO: Our final speaker of the morning, Dr. Pat Wojtkiewicz, earned his Ph.D. in cellular and molecular biology from Tulane University. He is currently the director of the Shreveport Laboratory, North Louisiana Crime Laboratory System, in Shreveport, Louisiana.
This morning's talk will be about the application of a single robotics platform to the combined purification, quantitation, and amplification of DNA in forensic samples.
DR. WOJTKIEWICZ: First of all, I thank NIJ for allowing me to talk, and also for supporting this particular research platform. Wojtkiewicz: Slide 1
This is a type of research that casework analysts would find important because one of the goals is to increase the casework capacity of a laboratory by using a robotics platform to analyze less complex samples and forensic casework samples. These are the samples that probably have lots of DNA, but yet still consume the time of the analyst. Another benefit of this research is that it will free the analyst to perform the more complex tasks on the more difficult samples, for example, the ones in which you have very small blood stains. They're going to have more free time to look at evidence and to interpret more difficult materials, mixtures, and things like that. Wojtkiewicz: Slide 2
When we first got into this research and looked at the type of robotics platform, we had a set of general specifications. One, we wanted to transfer liquids without cross-contaminating samples. As you know, robotics have been involved in DNA analysis for many years. Laboratories that look at convicted offender samples are dealing with very large amount of samples. These samples, however, can be examined repeatedly, and contamination is probably not a gigantic problem. But with forensic samples, sample-to-sample contamination is a major concern. Wojtkiewicz: Slide 3
We also wanted to use the Qiagen DNA purification protocols. Our laboratory uses organic purification and Qiagen. We wanted to use something that we were familiar with and had validated in our laboratory.
We also had an ambitious goal. We wanted a single deck to perform and quantify the DNA; that is, we wanted to use the quantified amounts to set up and run the polymerase chain reaction (PCR) on the same deck for both Profiler Plus and COfiler. We wanted to do this work or these steps without any or with as little user input as possible. In other words, what you do is you put the samples on the deck, load it with reagents, go home, and return the next morning to an amplified product.
The optimal process specification would be to perform all steps without user intervention. We did not look—we were not looking—at high throughput. We're not a large laboratory. We can't put hundreds of samples on the deck everyday, maybe 40 or 50 samples or so every other day. At four or five analysts, we don't have enough analysts to feed an instrument that's high throughput. We just want something that would take a certain amount of analytical time away from the DNA analyst. Wojtkiewicz: Slide 4
Basically, we had a process that could go at somewhat slow speed. The robot could be a slower speed robot. Less than 48 samples per run is all we needed. Again, we wouldn't have the analysts to feed a larger quantity.
There were forensic sample considerations. When you're working with forensic samples in our laboratory, our policy is that you have only one sample tube open at a time. We may be transferring from a sample tube to an empty tube, but for the most part we basically have only one sample tube open at a time.
We also have different types of samples: blood samples, semen samples, saliva samples, or hair samples. All of these can be processed through the Qiagen process or through the organic process, so we were looking for something that would work with all types of samples. And of course, the big concern with forensic samples is cross-contamination.
If you look at overall sample processing flow, the colors that are in red are those handled by the analyst. The steps in blue are automated by either a robotics system or software. Wojtkiewicz: Slide 5
First you have your initial sample processing. That's the person who cuts out the sample and starts the digestion. Then you have purification. This would be part of the organic purification or the Qiagen purification, where you do the organic in Centricons. Then you have your quantitation. For the most part, I think most of us do QuantiBlot or PCR amplification.
You then have another red area, where you have to take out the amplified product and transfer it to either gels or a capillary instrument. After running the gels, part of the analytical procedure is automated, as indicated by the light blue color. Then you have some automated gel analysis, and at the end, you have the analyst intervene again to do some interpretation.
One of the things we want to look at or develop in this robotics system is entering every forensic sample into the automated process. That means if you have a hair digest that you've started with a digest buffer, it can enter the process. Blood stains, saliva, epithelial fractions, and sperm fractions can all enter the initial sample processing and in fact, can enter the robotic processing of the deck that we set up. Wojtkiewicz: Slide 6
You would not want to run very sensitive samples or limited samples on the robot. However, when you're testing a lot of samples (e.g., blood stains or cigarette butts), there are many times that you don't mind repeating some of it or doing it again because something adverse happened. You still have something to go back and look at.
Again, part of this robotic process is your basic Qiagen purification, which is the extraction purification that goes through the AL buffer, the ethanol, the AW1 and AW2 washes, and finally the TE-4 buffer.
This is the robotics platform. First off, we have a single pipette filter tip and also the NCC opener that uncaps the tube, takes the sample out, recaps it, and then moves the liquid from one part to another. On the deck, for the most part during all analyses, your sample tubes and your PCR plates are always capped, but there may be instances where they may not be capped. Wojtkiewicz: Slide 7
You also have a vacuum system. Qiagen has a centrifugation system for their robot kit, but you can also use a vacuum system to pull it through. To have total nonuser intervention, you have to use the vacuum system.
To quantify, we used a fluorometer-luminometer system, so you can use things like PicoGreen and Aliquant as your quantitation method.
We also have a plate stacker. Our robotics deck was getting a little crowded, so you have to have areas where you can move your plates around. You can also transfer plates in and out of there because typically you're going to have a limited number of spots on the deck to stack your plates.
We also have an on-deck thermocycler. The robot can take samples or plates from the luminometer and go to the thermocycler. Ultimately, the robot has all of the materials to do all the steps on the deck.
We've had the robot for a little over a year now. One of the things that we had to do was program it. Everybody here knows how to pick up tubes, how to stick the pipette in below the liquid level, pull it out, and so on. Well, the robot doesn't know all of this. You have to teach it first. You have to teach it where to go and that takes a fair amount of time.
Many times robot systems are sold as turnkey systems. You come in, stick your plates in, and it runs. The manufacturer told us that ours was uniquely designed, so we basically had to program everything, like how far the pipette tip goes into the tube, either all the way to the bottom or just under the liquid. Wojtkiewicz: Slide 8
We also had to deal with things like pipette speed. Some of the liquids that we use are kind of viscous. Well, you can't use a really fast pipette speed on that. Most of you have done pipetting, so you already know this and have adapted yourself to that. The robot doesn't know or do that. You have to program it.
Are there ways to make the deck location a little more efficient? We've actually moved some of the deck around to make our reagents more efficient to increase the capacity of pipette tip. We've written the scripts for it, or the macros, so that everything can be done.
We've also looked at the different types of recovery buffers. Qiagen uses AE buffer. We wanted to look at a different buffer. Qiagen AE buffer has problems on amplification inhibition if you use a very large amount of it. We looked at water and TE-4 and decided on TE-4.
We looked at recovery efficiency. We assumed that the organic extraction gives you the best recovery you're going to get or 100 percent. We looked at manual Qiagen and we got about 60 percent of the recovery than we did with PCIA (phenol/chloroform/isoamyl alcohol). In automated Qiagen, we got a little bit less recovery but that's probably because of going to the vacuum system.
But remember, we're working with samples that we know have a lot of DNA. Samples that have very limited in DNA would not be the types of samples processed here.
Also, in the future, we would like to explore methods that enhance recovery with lower amounts of DNA. In fact, Qiagen just announced that they have a method or they're doing a carrier system to increase the recovery of low samples. We're certainly going to look at that because that was in our initial grant application.
Looking at the sample processing flow, we've written the script of the full procedure (extracting, purifying, and quantifying the DNA; setting up the amplifications; and entering samples part way through) in modular form.
A sample that you've already extracted and purified in TE-4 can be put on the deck, and the robot can quantify it and set up the amplification for you. Or if you've already quantified the samples and you want to set up 40 samples with PCR, just put them on the deck and enter the quantitation amounts and the robot will set up the PCR. When done, you will have a plate with Profiler Plus and COfiler amplified samples. It just depends on where you want to put the robot in the process. It can go in any way.
We've actually run samples at all three processes (quantitation, amplification, and PCR) all the way through. Just put in the sample sheet and it goes through.
For quantitation, it was quite easy to set up the PicoGreen quantitation. But we really wanted to look at samples. We had to quantify samples that went from 0.05 to 5 nanograms per microliter. Five nanograms per microliter is pretty much the upper limit, and 0.05 microliters is about the minimum that we would use for our samples. We know that if we amplify 0.5 nanograms, we're going to get DNA profiles for our validation, because we are basically trying to target 1 nanogram per reaction. For out-of-range values (below 0.1 nanogram per microliter or greater than 5), we use a software patch to readjust the quantities so that it would set up the amplification. Wojtkiewicz: Slide 9 and 10
We've been using blood stain cards, which are about a quarter-inch hole punch, or buccal swabs, which are about one-fourth to one-fifth of an Omniswab, and we're getting plenty of DNA with low samples. Low samples are mainly what we're working with right now, especially when looking at precision, sensitivity, setting up the robot, optimizing its movement, and testing the different buffers.
One of the things that we have set up, done, and are looking forward to in the future is using PicoGreen with a fluorometer-luminometer, but you also can use the AluQuant system. We envision using quantitative PCR and amplifying some type of area on there. Wojtkiewicz: Slide 11
We have used PicoGreen quite successfully so far. Some of the research has just been organizing the deck and moving plates around. We don't make our own standards. We simply put a standard on the deck and let the robot make the standard curve. It makes all the standard dilutions, puts them on the plate, and does the quantitation. Right now with PicoGreen, we've had a standard curve that's quite linear from 10 to 1,000 nanograms, which is basically the range that we're looking at—0.05 nanograms to 5 nanograms per microliter. Wojtkiewicz: Slide 12
We compared the results that we got from the PicoGreen with what we got from QuantiBlot. Our results with the PicoGreen are slightly greater. It could be the fact that sometimes the buccal swabs may have bacterial DNA that's been quantified by the PicoGreen and would not be quantified by QuantiBlot. We've taken that into account and basically looked at how much DNA we want to amplify. We've seen PicoGreen give us close quantitation. I don't want you to think that it's like twice as much; it may be 5 or 10 percent higher quantitation than what we're seeing with QuantiBlot.
The robot takes the quantitation values and uses those values to set up PCR amplification. We target 1 nanogram per reaction. The extracted DNA is put into the plate and recovered in that plate, and the robot prepares the dilutions right next to the plate. These dilutions are 1.1 nanograms per microliter—that is, a 25 microliter reaction gives you 1 amplified nanogram of DNA—and the robot just sets those dilutions up adjacent to the plate. Wojtkiewicz: Slide 13
You then can use those dilutions to set up another plate—another child/daughter plate, I think they call them—where you have Profiler Plus and COfiler all in the same amplification plate. We then can set up our controls and interim columns and go to the thermocycler to amplify it. Basically, it would take 9 or 10 hours for this to occur. Most of the time is spent in amplification, about 3 hours or so. Wojtkiewicz: Slide 14
We have tried to automate the process even more through Microsoft Excel software. We've created a script or a macro in Excel where we enter the data to set up the plate database and ABI (Applied Biosystems) I 377 sample sheet. You can export to Excel. We also intend to look at GeneMapper ID as another automation option for the final processing so that we can actually go all the way through the process with only a couple interventions, not on the robotics deck but only prior and after. Wojtkiewicz: Slide 15
How we look at processing the data or how we intend to look at processing the data. It goes into the initial sample processing or to control the samples. You also have the sample sheet that goes into setting up your 377 or 310 run. We use Excel, ABI data collection, and GeneMapper ID programs. Wojtkiewicz: Slide 16
But we also use a sample sheet to set up process controls of our samples, where we export the sample sheets and sample identification into the robotics system. Next, we use Excel to do that little patch that calculates the out-of-range values or quantitation values so that every sample will be correctly amplified, and then we can use the export of that sample sheet to go into ABI data processing or prepare an ABI or GeneMapper data collection sample sheet. The other automated processing would be through GeneMapper. We're trying to automate that process to where it will take the samples from data collection and go all the way through to allele calling.
As far as experimentation goes, we've looked at sensitivity, and we've taken samples or known quantities of DNA in digest buffer, diluted them, and have been fairly effective at identifying or being able to quantify samples that are 0.1 to 0.11 nanograms per microliter. With the Qiagen, once you get about 4 to 5 nanograms per microliter samples, you start seeing plateauing. That's basically because Qiagen is probably capturing the DNA, and it can only capture so much. It's a capture-and-release type process. However, it does show that the sample, when it's there, is collected in a fairly regular manner and it's consistent with what you would expect to see about the dilution. In the end, you're not losing any DNA during the process. Wojtkiewicz: Slide 17
We have seen several instances of outliers; that is, samples that came up with exceedingly low quantities even when we would expect much higher quantities. We've also seen this in the manual Qiagen. We suspect that there are just a certain number of Qiagen columns that have a problem. They don't capture the DNA nor do they collect it, because we notice that about 1 out of every 20 samples of our manual Qiagen also come up without any DNA. We're seeing the same thing here on our deck. We don't think it's a robotics problem. We think that it's a problem in the Qiagen columns themselves.
We also looked at some previously analyzed and typed sperm fractions and tested them at all three stages of input. We knew the quantity of DNA in there, and we simply set the sperm fraction on the platform. There were 11 of them. We manually entered the concentrations, and then had the robot set up the purification, make the dilutions, and set up the amplifications. All of those amplified quite well and came through the system and could be analyzed or read. Of course, they all matched what we already had known. Wojtkiewicz: Slide 18
In summary, we were able to use a single robotics instrument platform to purify, quantify, and amplify forensic samples. Basically, it means we input the samples, started the digestion of the samples, put them on the deck, and the deck handled everything else. When it was done, all we had to do was take out the plate that had the amplified products (the Profiler Plus and COfiler), and then, of course, we had to load the gels. The samples can be used without user intervention. Again, like I said, you put the samples on the deck and when the process is done, you have amplified product. Basically, the analyst performs the initial sample processing and genetic typing preparation and interprets the results. Everything in between is handled by the robot. Wojtkiewicz: Slide 19
Now, certainly we've had some problems along the way. For example, we ran through a whole set of samples twice but they didn't work the second time. Then after a lengthy search, we realized that we hadn't programmed the pipette tip to go deep enough. So it always worked the first time, but the second time through, the pipette tip wasn't going quite far enough. So, little things like that can go wrong, but we just modified the script and did it again. We also found that the tube opener does have a lifespan. Once it gets to the end of its lifespan, it doesn't like to open tubes any more. It will simply get there and click and continuously try to open them.
So there's little things like that that we've been discovering along the way, for example, problems with the Qiagen columns not collecting the amount of DNA and contamination. Initially, we did have a problem with cross-sample contamination. Well, we found out that we had our vacuum system set too strong. So we reduced the vacuum system but still had a little contamination. Eventually we found out that you have to give a few of those Qiagen columns a little air to breathe. You can't cover all the columns or put liquid in all the columns. It tends to make a little spray at the bottom. Since we learned to leave a few columns unused, we have not had any incidents of contamination.
Again, we've had to work through some problems, but overall we've had quite a bit of success with the samples. We're looking to go to reference samples fairly soon. We think that the validation works well for them. The PicoGreen is good because it's not human specific and you do not have to have human-specific DNA quantitation on reference samples. We think that we can certainly start going into our reference samples using them pretty soon.
First of all, I'd like to acknowledge NIJ for inviting me here and supporting this research. Michele Collins has done a lot of work with this. She is one of our analysts and, unfortunately, she's about to leave the laboratory. We also had a student intern, Corina Blackburn, who worked quite extensively this spring. She ran a lot of samples, set things up, and basically did the day-to-day work on these types of samples. Wojtkiewicz: Slide 20
MR. TAMBASCO: Thank you, Dr. Wojtkiewicz.