Thank you, ladies and gentlemen. Please be seated. Dr. Cotton, would you resume the witness stand, please. The record should reflect we've been rejoined by all the members of our jury panel. Dr. Cotton is again on the witness stand undergoing direct examination by Mr. Clarke. Mr. Clarke, you may continue.
Dr. Cotton, I believe we left off and you had described a first visit made by experts from the Defense in this case?
You also described the fact that he in fact actually performed some cutting of some samples at that time?
All right. I would like to return to that in a little more detail later. But as far as these cuttings, was his role limited, as far as the cuttings are concerned, to actually physically cutting a portion of those pieces of evidence?
And did those--the entire evidence including both sides of the cuttings so to speak remain in your custody?
Without getting into the date, Dr. Cotton, do you recall who was present during that visit, who actually made the visit?
He came to the laboratory to basically look at the laboratory layout. We did that during the visit and he was--the mission was to examine documents that we hold at cellmark that he or the Defense wished that he look at.
All right. With respect to the types of cases your laboratory takes, you described the fact that they involve human identification or forensic casework?
As far as other cases, does your laboratory have any contracts or have you had any contracts with the federal government?
We had a contract with the Department of Defense. I don't know that that--the contract is not still working. I don't know what the right term for that is, but it's done. And the contract was for doing analysis for desert storm casualties.
Now, approximately how many cases does your laboratory perform work on in the course of a year?
Oh, I'm sorry. That 4- to 500 would be forensic cases only and the paternity cases would be, oh, somewhere between a thousand and 1500.
Basically, it's the set of procedures that are worked alongside your normal work to make sure that the equipment is functioning and to ensure that the quality of the work that comes out of the laboratory is always high.
Umm, she keeps--well, she does a whole lot of different things. She--it might be a little easier just to sort of say what happens in a quality assurance program.
But, for example, it includes being able to track and when reagents are made. That is things come into the lab and they're used to make solutions. So there are records of those solutions. There are many different things in the labs that have to be at constant temperature. So all the refrigerators, freezers water baths and so on, all those temperatures are maintained. She's also responsible for administering and looking at proficiency tests. She does audits of the laboratory. That is, every month or two months, she will walk through the laboratory unannounced and check out to make sure that all the equipment is in order. Anything that should have been calibrated or adjusted on a routine basis, the records are in place that those calibrations or adjustments have been done. And I guess that's the general idea of what she does. And now I don't remember where I was supposed to go from there.
Just in terms of quality assurance, this person performs this function of overseeing, quality assurance as you described it?
She oversees--yes. She oversees it. That doesn't mean she goes and does all the temperature checks, but she oversees the entire program.
Dr. Cotton, if you would, would you allow Mr. Clarke to finish asking the question before you start your answer?
What steps do you take to ensure that when you receive, for instance, a set of evidence samples and let's say one or two known samples, to ensure that the chain of custody or the integrity of that evidence remains proper?
The samples are assigned to a specific analyst. The analyst logs in each of the samples on a form. That means he or she writes down a specific description of the items that are received, how they were received, from whom they were received. We usually get a letter with the samples that says who we are allowed to contact regarding results, who we're going to send the bill to and so on. The analyst has specific areas within a locked evidence room where he or she stores the materials. The analyst has custody of the materials the entire time they're at cellmark and during the testing, keeps detailed records of where the samples are, at what stage in the testing. That is, she doesn't write down where they are because they would be in her area of the refrigerator or freezer, but she or he would write down they're in the freezer, they're in the refrigerator if that was important to keep track. And there are a lot of forms that get filled out during the testing that document how the testing was performed, and then at the end of the testing after the report is written, that person then returns the evidence or anything remaining from the evidence to the submitter.
You've described earlier that two individuals who actually performed testing in this case were Julie Cooper and--I'm sorry. Did you mention the second individual?
Julie originally received the samples in this case, and sometime after she started the samples, she accepted a job at the Maryland State DNA laboratory where she's now a supervisor. And so because she was leaving cellmark, she transferred custody of all of the materials at that time to Paula Yates who continued on with the testing until it was completed.
Can you briefly describe each of their qualifications at the time of their testing?
Both of them had worked for cellmark for about six years. Paula is the supervisor of the forensic group, has a bachelor of science degree from the university of Maryland and about six years experience of doing forensic testing. Julie has a masters degree in forensic science from George Washington university. She also had about six years of experience in forensic testing.
Had you prior to this case previously worked with both of those individuals, Julie Cooper and Paula Yates?
And is that something--well, let me rephrase that. Did you in fact work with them on a number of cases prior to this case?
As far as the testing techniques that are used in the laboratory, do you have in place protocols about how tests are to be conducted?
Basically, it's sort of like a recipe book. It lists all the steps on how to do the test in--in a lot of detail.
Yes. There's a separate probe--there's one book basically for RFLP and another book for PCR testing.
Is it common in science to have protocols describing how to conduct scientific tests?
As far as these protocols, is it something--for instance, for either RFLP typing or PCR base typing--that is relatively short or is it detailed and extensive?
I've seen protocols that are relatively short and I've seen them that are detailed and extensive, and ours happens to be detailed and extensive.
Do you have any role in, for instance, either the creation or changes that are made to these written protocols?
As laboratory director, my signature is required for final approval for any change in the procedure and my signature is required yearly on the procedure as evidence of the yearly review. So I participate in that yearly review with the quality assurance coordinator, and if we decide to make a change or an addition to the protocol, someone else certainly might write that and many people may look at it, but it requires my signature as laboratory director to be put into place.
Umm, many of the changes are simply additions. That is, we might add--for example, we were doing DQ-alpha and then we wanted to do the poly-marker testing. So we had to write protocol that now included the poly-marker testing. But the other thing is that you might go to a scientific meeting or read something in the literature where someone had discussed a procedure that gave very good results. You would then check that out in your laboratory, and then if it was getting better results than your current procedure, you would go ahead and change your protocol to reflect the improvement.
As far as these protocols, do they start at a particular portion of the testing process and end at another one? Does that question make any sense?
From--it starts out how do you log in the evidence. How do you properly receive the evidence is where it starts.
Are individuals who are working in the laboratory such as Julie Cooper and Paula Yates required to follow these protocols?
What happens if something comes up in the course of testing that isn't covered by the protocols? What happens?
Umm, generally, first you would make some notations about that in the case notes and then you might consult with someone else. If there's something--if we think that something should be done that's unusual, that's outside the protocol, then we'll make notations about that, and that generally would require approval of one of the Ph.D. staff.
Were your protocols reviewed by the Asclad inspection team prior your receiving their accreditation?
Now, you've described the fact that records are kept in the course of receiving evidence through the various testing processes; is that right?
Well, my role in the case is to get all the records and including the results and a draft report and review the records, review the data and review the report and participate in signing that report with the person who actually did the work with their own hands. So as part of my every-day job, I'm going through case records, reading them, checking them, making sure that they're all in order so that I can sign the case when it's done.
To be able to describe results in a given case, how can you do that without actually sitting there and watching every step of the testing process?
When you're done with the test, you have some data. You have some information. So what I'm doing is participating with the analyst in interpretation of that data. And I have not sat and watched--I mean, if I was going to sit and watch, I might as well do it myself. I have not sat and watched every step of the test.
As far as your ability to offer an opinion, do you review all of the material in a given case prior to offering any opinions about it?
In either case, I would be reviewing all of the data and most of the documentation in the folder.
Okay. I'd like to turn your attention to an area that's commonly referred to as population frequencies. Is that term familiar to you?
It's the expression of saying for a particular genetic characteristic within a population or large group of people how often would you expect to find that characteristic.
For instance, would population frequencies enter into a description of approximately how many people have the ABO type, blood group type O for instance?
As far as population frequencies, could you describe, please, the education and training you received in the area of frequencies?
The--my exposure to this particular area of biology has mainly been in the seven years I've been at cellmark because we deal with defining frequencies for the genetic markers we use every day. So I have no specific training outside of cellmark. We have a person on our staff who has a Ph.D. in this particular area. And as a part of my job, I--and when I go to scientific meetings, I'm both hearing lectures in this area and I make it a point to read the literature in this area as well.
Well, in terms of doing a case, if you have samples that are consistent with some known person and this is part of your report, then as part of your report, if there is sufficient data, you're saying how often you might see this particular set of genetic characteristics. So in all the report--not all, but many of the reports that go out involve the calculation of frequencies or particular sets of genetic characteristics.
Do you consult with experts in the area of--well, let me rephrase that question. Is there a particular area of science that deals with, in particular, the area of populations and frequencies of genetic marker characteristics?
Do you and have you as a result of your employment at cellmark since--I'm sorry--what year?
Have you consulted with population genetics experts about this area of population frequencies?
And I believe you mentioned that you had an individual in your laboratory who--whose field of specialty is population genetics?
As far as work conducted in your laboratory, do you then use population frequency data as a means of describing the significance of your results?
To your knowledge, has population frequency data been used for a substantial period of time?
Well, it's certainly been used in DNA typing since laboratories were initially engaged in that. But in terms of formulating information about serological results, the same kinds of calculations and information is usually provided. So if you have a set of information, then it includes blood types and PGM or however many markers you've done, you can do a calculation to say this particular set of markers would occur in some percentage of a particular population.
Is that related to, is it similar, different from population frequency use in DNA typing?
Your Honor, I would object and I would ask that I conduct very brief voir dire on her expertise in this particular area.
Actually, your Honor, would it be possible to have that reread because I can't recall it exactly.
As far as this use of population frequency data, has it been in use for a substantial period of time?
As far as its use, that is, the use of population frequencies in casework, when did you begin using population frequency data in terms of reporting it in actual cases?
Okay. There's going to be a little bit of time--there's a time frame in here where I can't be particularly accurate about the answer. I came to cellmark in 1988, and sometime during that year, I started being involved in forensic casework. At the time that I started being involved in forensic casework, we were in the process of developing population data that would allow us to do that. I think that we were all providing some frequencies before that based on population data from the other cellmark laboratory, but I'm not absolutely certain.
The point at which I would have been--become involved would have been the point at which I began signing forensic casework, and I was also involved in setting up how the population samples at cellmark would be analyzed and reported. So at that point, I became involved in discussing, understanding and generating population frequencies as they relate to DNA typing.
Do you have any role in supervising the actual reporting of frequencies in your laboratory, that is in casework?
In terms of signing a particular case, both people who are signing that case have a role in generating and reporting that information. So in my role in signing cases, I am responsible for the information that's generated for that particular case and I also have overall responsibility for the scientific information that comes out of the laboratory. So under that umbrella, I guess I would have some responsibility as well.
As far as your previous testimony as an expert in DNA typing, have you reported population frequencies during that testimony?
And I'm sorry. How many times had you said approximately you've testified previously?
And amongst that 90, is there any way you can put a figure or a rough estimate of the percentage of the time you report population frequency data?
I couldn't give you an exact figure since I'm not even giving you an exact figure of how many times I testified. But for most of the cases, probably 80 or 90 percent of the cases in which testimony is provided in Court, we are providing some population genetic frequency for a particular set of markers.
The written protocols you have in place in the laboratory, do they describe the reporting of population frequencies?
The calculations, although they're done by computer, can also be done by hand, and the protocol simply describes how to go about entering that data for--and that's for RFLP. For PCR results, we do not have a computer program to do the calculations for us. So the calculations are done by hand. And when you review a case, in my role in reviewing a case, those calculations have been done by the analyst, I'm checking them and then we're both signing the report.
When you say that those are done by hand, do you mean by using a calculator or really by hand?
No. I mean by writing the individual frequencies down so that they--we have a permanent record of them and then by using a calculator, to multiply them together.
All right. With regard to this concept of population frequencies, how do you calculate them for a given type, for instance, at a given genetic marker?
With regard to this area of population frequencies, is there a term called "database"?
It's basically a sample of a population. If you--obviously, you can't sample all the people in the country or you couldn't even sample all the people in a given state. So you need to take a small subset of the total population and then regard that as your sample. It's similar to taking a poll for television viewing. Nobody calls up everybody who's watched television on a particular night. They take a sample of the people who have watched television on a particular night and make some assessment of the larger sample based on the small--the larger population based on the small sample.
And how does that play a role in your ability to be able to estimate how common or how rare a set of characteristics is?
Well, I think you've just actually said the important word. That is "estimate." If I have a sample of 200 individuals, for example, and I do an RFLP test on all those 200 individuals, I've now looked at--for any given genetic locus, I've looked at 400 chromosomes. So I have 400 pieces of information. And from that, I can ask the question if I have a DNA band that's 5,000 base pairs long for this particular genetic locus, how often in my population sample do I find a person with this genetic characteristic, a DNA band at 5,000 base pairs.
Does that involve then the use of more than one database for a particular genetic marker?
Our databases at cellmark are divided into racial or ethnic groups. You could maintain any number of databases in our laboratory. We have three. They are popu-- and I guess for the moment, I'll just speak about RFLP. But the situation for PCR is similar, and they are divided into African Americans, Caucasians and Hispanics.
With regard to the area of population frequencies, are you familiar with other DNA laboratories and their methods of calculating these estimations of the--how common or how rare characteristics are?
Both from reading the literature, talking to other scientists and hearing information presented at scientific meetings.
What about scientific literature? Does it play any role in your knowledge of the area of population frequency data?
Well, there's actually--at this point, there wasn't--this was not the situation in 1988 for example. But at this point, there is a lot of information in the literature about population data based on DNA typing. So anyone who is interested or involved in doing DNA typing analysis either for paternity or for forensics would be interested and need to keep up with this literature.
What types of literature do you read in that area other than the general topics obviously? But are there particular resources that you have in terms of particular journals and so forth?
There are two journals in which--which would include most of the papers, certainly not all. One would be the journal of forensic science and the other would be the American journal of human genetics. Now, there are a few papers that are somewhere else. There's a couple in science and--anyway, they're here and there. But for the most part, the two journals have covered more of them than maybe others, some other journals.
If you could describe, how do those journals, how closely do they deal with this actual concept of population frequencies and reporting results following DNA typing?
All right. With respect to your review of scientific literature regarding population frequencies, can you just tell us, as far as the literature you've described, does it address the reporting or the estimation of frequency data in DNA typing cases?
It doesn't so much address the reporting. What the literature is addressing is how the frequencies are calculated and--and how individual frequencies are determined and how populations differ. That is, are there large or small differences between African Americans and Hispanics or Caucasians and Hispanics in how frequently these particular characteristics occur.
Do you regularly speak with population geneticists about the methods used in your laboratory? And you've described the one on staff. Are there other individuals?
Uh, we have a particular person who's a population geneticist who we speak with on a regular basis. And that would be Dr.--
Dr. Bruce weir is the Professor at north Carolina State university. He is a population geneticist and very highly regarded in this area.
In terms of these consultations with Dr. Weir, has he reviewed material at your laboratory that you utilized to report population frequency data?
As a result of that, have you incorporated any suggestions, comments, et cetera, that he may have about the methods of reporting data in your laboratory?
The basic answer to the question is yes. Most of his input is in some data that's being currently developed. He's also looked at much older data that was developed by our laboratory, but he hasn't--in the older data, he has not made any suggestions that resulted in large change or really any changes in procedure to speak of.
And that was after his having--when you say no changes, that was after his having looked at these procedures in place in your laboratory?
All right. Is there a way you can describe the basic method by which population frequencies are report--are actually calculated?
Would it assist you to use a drawing pad or would you prefer to do it just by description?
Thank you. Dr. Cotton, if you would--and this will be People's exhibit 255. Do you have a title that would describe this particular chart? Population frequencies? Is that--
Population data? All right. Now, if you could use that diagram to demonstrate the basic method by which population frequencies are determined to describe how common or how rare a particular genetic marker type is.
With regard to your laboratory and how you calculate an estimate of how common or rare a characteristic is, how do you do that?
You are going to do that based on a population sample that's either been developed in your laboratory or somewhere else. In our case, our population samples for RFLP were developed in our laboratory. So if you go back to the explanation yesterday, what you're doing is, you're taking a sample of people and you're doing the RFLP test on all of those people. So what I'm going to do on the diagram is sort of show you what--give you an example of what a sample of people looks like and then show you how that would be used to estimate a frequency. So we're going a make a mini population on the diagram.
Now, let me stop you if I can, Dr. Cotton. You've written on this chart labeled "population data" what appears to be the numbers 1, 2, 3, 4 and 5?
And then below each of those numbers, you've written in what appear to be two bands? Is that what those are intended to depict?
Right. In other words, I'm trying to show you a diagram of an autorad with five samples.
Okay. In my five samples--because I have two bands for each sample. That is for each of these people, one for mother, one for father--I've got 10 bands altogether.
Actually, would it help to draw basically lines between these lanes so that at least on this diagram--
--it would be easier to tell? And maybe in a different color also. You can start as high as the numbers themselves that label the samples.
Okay. Now, let's suppose that we have yet one more sample. We'll call it A. I guess I should make-- and let's say-- could I have one more color, maybe black or something?
Mrs. Robertson, do we have a black marks-a-lot? I think I have one back by my microwave. That's just a highlighter, doctor. That will fade.
We'll go back to my very simple example. And let's remember that where a band is, looking from the top to the bottom, is related to how big it is. And so let's say that we've determined that this band is in sample a is 5,000 base pairs. And so the question then is, well, how often would we see another person with this particular band. And although every time you do this, you're not going to go look at a whole set of x-ray films, you have all the data in a computer from those x-ray films, and the question you're saying is, "I have a band at 5,000 here in my population sample. How many other people do I see that have a band there?" So let's say that this band was also 5,000 and this band was also 5,000 and this one was also 5,000.
I'm saying the length of DNA that formed this band was 5,000 base pairs long, using base pairs as a measure of length.
And I've made this example very simple because I've made each of these bands be exactly the same size as this one. And that's not really necessarily--they wouldn't be necessarily that close. So out of the 10 bands that I see, three in my sample are 5,000 base pairs. So three out of 10 people, three out of 10 chromosomes would have a band at 5,000. So .3 would be the frequency for this band based on my limited sample.
Then you would go through the same exercise for the second band in that sample. And say this was--I didn't give this any counterparts. That's not so good. I'll tell you what. Let's get rid of this one and make it be here. Suppose this size was determined to be 2,000 and this size was also 2,000. And in this case, for this band, we had one out of 10 chromosomes that had this size. So the frequency for this band would be .1. The frequency for this combination then would be two times the product, .3 times .1.
All right. With regard to the step you just did in terms of multiplying, what's the reason for that? Why do you multiply?
As far as--and let's talk about the databases that you have in the laboratory. You've drawn a diagram with basically five different persons' DNA on this particular chart?
The databases in the laboratory range in size from about a hundred and fifty to somewhat over 300. And the reason I'm saying "range" is that for one genetic locus, we may have--and for one racial group, this number of people that we've sampled may be different than for another genetic locus or another racial group. So that's the range of size--sizes of the samples that have been done in our laboratory.
Is that enough of a number of samples to look at to be able to estimate how rare something is?
With regard to this examination of databases--and you've used the example of what? Results of an RFLP test here?
How are these databases created in terms of, how do you determine out of these 150 or more samples the various sizes of the various bands?
It's done in exactly the same way the casework analysis is done. You extract the DNA. You cut it with an enzyme. You run it on a gel. You do a southern blot, you get an autoradiograph and you size the--you estimate the sizes of the bands and the autoradiograph with computer imaging system exactly as you do a regular case.
Is this the same--this process you've described up to now--the same or different from those methods used in serological cases like methods that typed PGM four years previous?
This principal of looking at databases and determining relevant rarity or how common characteristics are.
That would be similar to how this--those calculations are done for serological markers.
As to serological markers, once you've determine let's say an approximation of how common or rare a PGM type is, are there, using serological techniques, similar estimations made of the same sample of another genetic marker like ESD or GLO?
To your knowledge, how similar is the method of calculating frequencies in DNA typing to previous methods that have been in use for longer periods of time?
All right. With respect to these fragments that you've described--and let's go back to the chart you've created. And you show 5,000 for what is it? Four of the particular markers and then 2,000 for two of them?
What role do the other fragments play that you have not at this point labeled an approximate size to as far as determining from databases how common or rare something is?
So if you want to calculate how often you see something, it's going to be some percent of the total. So we have 10 total, three is some 30 percent of that.
And as far as your own training in terms of reading the literature and performing these calculations in your laboratory, is what you've presented simply a simplified version of what you have reported in cases for a number of years?
What happens when you look at additional markers as well? In fact, let me back up. Let me rephrase that. The diagram that you've just created relates to one genetic location?
Do you look at more than one genetic location normally when using the RFLP technique?
Normally you would look at anywhere between three and five. Now, you could look at more. Our laboratory I should say--let me restate that. Our laboratory has the capability of looking at five.
These locations are what? A different--well, what's the technical term for the term "location"? Do you have a word that you actually use in DNA typing?
Yes. It's--it's the chromosomal locus. So the word that you'll hear is "locus." When you say, "I have a DNA probe or a genetic location," what you're really saying is the locus, which means the address on the chromosomes for this piece of DNA.
Now, Dr. Cotton, referring to this diagram again and using the example of this one genetic marker location, in your training--and let's go back to the education you received in terms of your formal education. Did any of these concepts of population genetics arise?
Well, to tell you the truth, it's been so long since I was in undergraduate school, I really couldn't tell you what course they were in. Most of the principles that are used in DNA typing that relate to population genetics, you can find a pretty good discussion for in any college genetics textbook. Now, that doesn't mean the area is simple. There are complicating things that--that when an analyst such as Bruce weir looks at a set of data, he is doing something more complicated than what's in a college genetic textbook. But the principles that are involved in doing these calculations are generally outlined at that level. That's where I would have been first exposed to them.
Okay. And then where next in chronological order would you have then been exposed to these same concepts?
And can you describe in a little more detail how you came about to be able to calculate these frequencies?
Basically it was clear that we needed to be able to make an estimate for a particular genetic--for a particular DNA type or a DNA profile. It was clear that part of the information we needed to provide was how common or rare that profile would be. Umm, with assistance from the cellmark lab in the U.K. And discussions with the staff at cellmark, the group of us that were there at the time sat up the procedures that we would use, gathered the population samples, did the analysis on the samples. About the time we were doing the analysis on the population samples, Dr. Foreman came to the lab. So it was really a combined effort, certainly not mine alone, to set up the procedures that we are now using for creating estimates for population frequencies. Now, what I'm--let me make something clear. The calculations are laid out. They make some assumptions. In fact, the particular assumption that the calculation makes is that what's inherited at one genetic location does not affect what's inherited at another genetic location. These sets of assumptions are not--they're not--it's not a formula that we devise. The formula that's used in these calculations was developed in the 1930's and is the same type of formula that's used in the calculations for serological markers. What we did in the laboratory was develop the population data and then the methods to sort of query that data. And really what I'm getting at is that when I have a band that's 5,000 base pairs, there can be some variation in that size if you run the same sample over and over again. So when we query the--when we go to the database, we've never looking for something that's exactly the size. We're looking for a range of sizes. And the methodologies that we use to determine those ranges were all developed at cellmark by the group of us who were working on that project at the time.
Now, you've described in sizing these samples just now the fact that it's an estimation; is that right?
As part of an estimation then, when you look through databases to see if the sample matches or not--well, let me rephrase that. When you look through a database to determine approximately how rare or how common a particular characteristic is, do you take any steps to make sure that you don't overstate the rarity of that characteristic?
The situation is that, when you run a DNA sample on an agarose gel, as I described yesterday, that procedure has limitations. You're getting an autorad, you're using a computer imaging system and you're estimating the size of the DNA that created a particular band. Even if I take the same sample and I run it 20 times or 50 times, the size estimates will not be identical. Let's go down and write underneath some--a little more realistic example.
Yes, I am. Let's say--and I'll put my new more realistic example in parenthesis. Let's say that this size came out to be 5,010 base pairs and the next one came out to be 5,030 and the one over here came out to be 4,982. None of these numbers are exactly 5,000. So in my little mini database here, if I ask the question, now assuming that the figures are the ones in parenthesis, if I said, how often do I see a person with a band at 5,000, I would look across here at these numbers and I will say, well, I hadn't seen one. So--in my 10 people, I hadn't seen anyone, and then you would have to make some assumption, well, if you look at more people, would you see one or, you know, you would have to make some calculation after that. So basically, I would very much underestimate this frequency if I look only for this exact size. But we know that there is imprecision in this gel system. It exists in every gel system that's like this that's run for RFLP typing. It's a feature of the system and it's important that if you're using this system, you understand that.
Is that a--I'm sorry, Dr. Cotton. Is that a feature--when you say it's a limitation, is that something unique to your laboratory?
No. What I'm trying to say is that any laboratory that's doing RFLP typing and running an agarose gel will have this same problem.
And when I say "problem," it's not an insurmountable problem, but it's something that you want to be aware of. So if I now say, okay, I recognize that there's some imprecision in this measurement and based on some work that I've done in the laboratory, I recognize that the range of sizes that I might get might go anywhere from a low of 4,980 to a high of 5,030. Now when I go to my database, I'm not going to look just for a size that's 5,000. I'll look for something in this range because I know or I've determined in the laboratory that this is the realistic size range that I could expect if I ran something over and over and over again.
Well, you could do it in a--do it just what I said. You take a sample and you run it over and over and over again and you see how much variation you get when you do that.
And we've done that and it's worked. Anyway, now when I go to the database, I'm looking at this window of sizes. And now based on that window of sizes and the numbers that I have in parenthesis, I would say I saw three people in my database whose sizes were in this range.
Does that have the effect then of ensuring that you don't overstate the rarity of characteristics that are similar or the same?
Why do you adopt that approach in terms of frequency data calculation in your laboratory?
Because that's the logical approach. You can't--if you know the system has a limitation, you have to work around that limitation.
Okay. You also described the existence of a formula for when you're looking at more than one genetic marker?
The formula that's used in these calculations was developed in the 1930's and is the same type of formula that's used in the calculations for serological markers.
If I said, how often do I see a person with a band at 5,000, I would look across here at these numbers and I will say, well, I hadn't seen one... you would very much underestimate this frequency if I look only for this exact size.
I think that's about the sixth time we've asked that question.
The particular assumption that the calculation makes is that what's inherited at one genetic location does not affect what's inherited at another genetic location.