Hi jimhuff,
A good overview of Y-Chromosome testing is found at:
http://www.dnanow.com/dnanow/botframe_files/grandparent.htm
Basically, Y-Chromosome tests are completed in cases where
male-lineage is a potential issue needing resolution. Because only men
have Y chromosomes, and because the same Y-Chromosome is passed from
Grandfather to Father to Son, you can ascertain whether or not a
great-grandfather is related to his great-grandson without any other
tests. Most DNA testing centers can confirm to 99.9%+ if the
Y-Chromosome is the same.
---
CIDR – FAQs
http://www.cidr.jhmi.edu/faq.html
CIDR is a lab that performs DNA testing: their main concern seems to
be that of quantity/quality issues surrounding buccal DNA samples.
---
forensic nurse – Advancing Forensic Science: A New Buccal DNA
Collector
http://www.forensicnursemag.com/articles/2a1feat3.html
This site suggests that between 10-30% of buccal samples need to be
redone for "one reason or another". It also calls into question
whether or not a sample is free of contiminates that could "bleed"
through the envelope that would contain the buccal swab.
---
Memorial Blood Centers
http://www.mbcm.org/parentage.html
This testing facility states that there is no difference between the
reliability (test-retest or repeat tests) of buccal cell DNA tests and
blood DNA tests. The only concern is based on the obtaining of DNA
from nucleated cells; that means, that a buccal swab may not get any
DNA on the swab—not that there would be "incorrect" DNA.
So, to answer your question, there is no difference between buccal DNA
tests and blood-based DNA tests: other than the considerations I’ve
mentioned already.
I hope this answers your question. Please do not rate this question
until you are satisfied with the answer. Should you need
clarification, use the ‘Request Clarification’ feature.
Thanks again!
Legolas-ga |
Request for Answer Clarification by
jimhuff-ga
on
25 Nov 2002 09:51 PST
These do not answer the question. I am aware of theory and buccal DNA
test background is not needed. During rational test methods
development, repeat measures are done. Someone has done the whole
buccal DNA procedure from swabbing to analysis repeatedly to achieve
"best standard method" normally called GLP for Good Laboratory
Practice. I want to see data, analysis and conclusions on experiments
that evolved the GLP. It is not (yet) believable to me there exists a
GLP method with zero error on repeats on any science measure. I have
read each of the references and they just do not relate to the
question.
|
Clarification of Answer by
legolas-ga
on
25 Nov 2002 10:00 PST
I will look for more information on your question and report back soon.
Thanks,
Legolas-ga
|
Clarification of Answer by
legolas-ga
on
25 Nov 2002 11:04 PST
I've attempted to locate some more 'clinical' studies: and am honestly
not having a lot of luck. I'd like to understand precisely what you'd
like to be able to narrow my search in the hopes of arriving at a
suitable answer for you.
First, some basic facts that I've found:
1) Buccal DNA is the EXACT SAME DNA as found in blood. DNA tests, at
reputable DNA labs achieve 99.999%+ accuracy. It *does not matter*
where the DNA comes from, as ALL DNA in your body is EXACTLY THE SAME.
2) Test-retest, or "repeated measures" is another way to write
"reliability". For a test to be "valid" it must be reliable. What that
means is that if a DNA test is a valid (or correct) way to prove or
deny a relationship between two people, it must be reliable. If a test
is not reliable, it is, by definition, not valid.
3) DNA tests are accepted as valid and reliable by courts. It is also
accepted as valid and reliable by the fact that it is 99.9999%+
accurate. There is a possibility that a test could be wrong, but, the
error is so slight, that a single retest (if the same result) would
put the error rate at such a small percentage, that errors would be
essentially impossible.
So, having said all that, my understand now of what you want (which
wasn't in the original question) is a summary of what is considered
"proper handling" of buccal swabs? Or, put another way, "How can I
argue that a buccal swab DNA result was not accurate?" Am I correct?
Please let me know,
Thanks!
Legolas-ga
|
Request for Answer Clarification by
jimhuff-ga
on
25 Nov 2002 13:33 PST
Well we are closer to an understanding of each other. I want four -
five complete DNA tests on one male, starting with the buccal swab
collection then all the wet chemistry involved in getting the samples
ready to inject into the machine, the injection, the results, analysis
and conclusions. The four - five tests should be done in the blind and
probably best they be spaced out during the day or week so that a
group ID is not possible either. From your comments it would seem we
should expect identical results with near zero error. I appreciate
that, but I would like to see actual data in support.
"How can I argue that a buccal swab DNA result was not accurate?" Am I
correct?
More accurately what is the probability that DNA Observations along
loci can be repeated in a second? third? or more tests, if you start
from the very first initial step every time?
Thanks for your interest.
|
Clarification of Answer by
legolas-ga
on
25 Nov 2002 14:04 PST
Hi jimhuff,
There is precious little information I can find on test-retest
reliability of DNA evidence. It seems that no one (or at least, not
many researchers) have devoted time to prove something that is already
accepted as 99.9999% accurate. However, I did find some other
references that you may find interesting:
The Evaluation of Forensic DNA Evidence
http://www.nap.edu/catalog/5141.html
This is a book that talks about studies conducted on the efficacy of
DNA tests. There is a link on the page to purchase the book at $34.36.
An executive summary and Table of Contents is available at:
http://www.nap.edu/readingroom/books/DNA/
Also, a list of "related" web pages appears at:
://www.google.com/search?sourceid=navclient&q=related:http%3A%2F%2Fwww%2Enap%2Eedu%2Freadingroom%2Fbooks%2FDNA%2F
Related web pages are those pages where the content subject area seems
to be similar to the subject page.
Specifically, you may be interested in:
Scientific Testimony: An Online Journal
http://www.scientific.org/
and
The National Institute of Justice: Publications and Products:
http://www.ojp.usdoj.gov/nij/fordocs2002.htm
The book seems like it is the best lead for the type of information
you are looking for. I did search on Medical Databases (like
http://www.pubmed.com ) for this topic area, and came up empty-handed.
I will keep looking, and report back if I find something else in my
travels, but, for now, this seems to be the best I can find.
One other good place you can use to get more references: is other
articles and books about this topic. If you buy the book I've listed
above, often there will be a lot of 'References' listed: those are
excellent places to look for other relevant articles.
Unfortunatly, because of the nature of the web, not all information is
freely available: and in this case, it certainly seems to be the case.
I'd look into the book in particular as the TOC seems to be
bang-on-target.
Thanks again!
Legolas-ga
|
Request for Answer Clarification by
jimhuff-ga
on
25 Nov 2002 20:19 PST
Thanks Legolas and we have a missunderstanding in that I am not
questioning the science of DNA per se, but simply put: What are the
probabilities you would get exact same results (counts/locus) if the
test were repeated multiple times from the start i.e. sample
collection. This question has little to do with DNA but rather how
well the entire procedure can be duplicated. I have also combed all
the sites I can think of and found nothing, which has led me here. In
my professional life I had lots to do with Analytical Test methods
development where the lack of the data-type I am seeking would be
irresponsible. My guess is that more likely in this case, it is known
and not publicized . . also irresponsible. I could buy the data by
submitting multiple samples, blind labeled, but at over $200 each, I
do not have that degree of anxiety yet. I submit, however, this should
be a subject of an academic paper. Again, thanks for your interest and
happy you are willing to give it more thought.
|
Request for Answer Clarification by
jimhuff-ga
on
25 Nov 2002 20:57 PST
I meant to add that I have travel plans from Nov 27 through November
30. If you have ideas or suggestions and I don't respond that is the
reason but I will respond on return.
|
Clarification of Answer by
legolas-ga
on
25 Nov 2002 21:03 PST
Hi jimhuff,
I think I understand what you want--but, I honestly believe it to be
more of a probabilities question than a GLP question. Here's why:
1) If we take for granted that DNA tests provide 99.9999% accuracy,
that means that on an given test, the probability of a correct answer
will be P=0.999999. An extremely high probability. The odds of getting
a correct answer on the second test is the same: P=0.999999 (for the
same reason that the previous coin toss doesn't help predict the next
coin toss). For definition sake, P = 1.0 is a guaranteed probability.
It is the "IT WILL HAPPEN" statement. P = 0.00000000 is a guaranteed
improbability. It is the "IT WILL NOT HAPPEN" statement.
2) Now, as our test samples increase, we want to see a higher
probability to account for measurement errors. In much the same way as
we'd want a higher P value when we do multiple statistical tests. P <
0.05 becomes P < 0.001, etc.. as we do more tests--as the tests might
introduce bias themselves. However, even if we want to see an accuracy
of 99.9999999999% at the fifth test, it becomes purely academic--as
the odds that all five tests were wrong was amazingly slight to begin
with!
So, then, what you REALLY want to know is... What is the probability
that all five tests could be wrong? It's kind of like a coin toss:
odds on one toss is 1/2. Odds on the second is 1/2 too. But, to get
both tosses to be heads, it's 1/2 * 1/2 = 1/4. So, 25% of the time,
flipping a coin twice will give you two heads in a row. As you add a
third or a fourth coin flip, the odds of getting heads on each
successive flip stays the same PER flip: but, becomes harder to do in
a row. So, 1/2 * 1/2 * 1/2 = 1/6 on three tosses, 1/8 on four. In DNA,
the odds can be expressed as:
0.9999 0.9999 0.9999 0.9999 0.9999 0.9995
------ * ------ * ------ * ------ * ------ = ------ = 99.95%
1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
Therefore, 99.95% is the probability that five tests in a row (using
the same methods and procedures), each giving 99.99% accuracy will
give the same results. It's also 99.9995% that five tests giving
99.9999% accuracy will give the same results. SO, if your test was at
99.9999% accuracy, only 5 time out of a 1,000,000 trials will result
in all five tests being wrong. Or, 1 time out of 200,000. One test is
more accurate than five due to the law of probabilities though as
demonstrated above.
I sincerely hope this helps you understand the issues you've raised.
Thanks so much,
Legolas-ga
|
Request for Answer Clarification by
jimhuff-ga
on
26 Nov 2002 08:21 PST
As best I can judge you have given the error associated with (in
essence) repeated injections of aliquots of a sample into a machine
over and again. I would not be surprised that error approaches zero,
given a well designed, calibrated machine run by a trained technician.
Well ahead of that (almost) final step there is a lot of wet
chemistry, extractions and purifications getting the sample down from
its human source to the final form for injection. It is hard to
believe all those steps can be repeated exactly time after time such
that the overall error is as low as you estimate for all counts along
the entire loci. That and machine error is what I am looking for. If
you want to quit, OK. I just do not think we are addressing the same
issues. Though it may be absurd to think of it happening, but until
tested for it, there is no data that speaks to the probability that
extraction or other procedures, does not change the count on one or
more locus and in a random fashion.
|
Clarification of Answer by
legolas-ga
on
26 Nov 2002 09:06 PST
I think what you are looking for is more of a consumer review of a
particular DNA lab. Kind of like, "Which film processors deliver the
best results?"
Nothing like this exists -- or at least doesn't exist publicly. I'm
sorry about that. I think what you are asking about it more likely to
be found in court cases where the defendant argues that the DNA was
improperly obtained, etc.. I provided a link to some sites like that
at:
Specifically, you may be interested in:
Scientific Testimony: An Online Journal
http://www.scientific.org/
It includes information on cases like the OJ Simpson case, etc.. What
you are looking for probably has more to do with the random human
error rate in DNA testing than anything else: but, that random human
error rate is why a DNA test can't say it's 100% accurate--only
99.9999%. That 0.0001% is the human error rate. DNA is a perfect
method of identification. If you could envision a completely
human-error and machine-error free DNA test, you could get 100%
accuracy. So, I think the answer you want is that it is 0.0001% error
rate from human and machine error. OR, 1 time out of 1,000,000 tests,
a human or machine error will produce an incorrect result.
Hope this clarifies the issue for you!
Thanks again,
Legolas-ga
|
Request for Answer Clarification by
jimhuff-ga
on
26 Nov 2002 10:21 PST
Yes I am asking about error rate associated with collection and any
sample prep techniques that could effect the test and I will read your
references. It has been an interesting discussion and thanks
|
Clarification of Answer by
legolas-ga
on
26 Nov 2002 10:28 PST
Also, just to call your attention to this link:
forensic nurse – Advancing Forensic Science: A New Buccal DNA
Collector
http://www.forensicnursemag.com/articles/2a1feat3.html
This site suggests that between 10-30% of buccal samples need to be
redone for "one reason or another". It also calls into question
whether or not a sample is free of contiminates that could "bleed"
through the envelope that would contain the buccal swab.
It may also have some more of what you are looking for...
Thanks again!
Legolas-ga
|
