DNA | Forensic DNA Investigation || Radcliffe Institute


-Welcome back. It’s a great crowd. It was great this
morning– still great now. I apologize that Ron
Sullivan is not here to introduce Greg Hampikian,
but I am thrilled to do so. You’ve got his bio
in your program. He is a professor of
biology and criminal justice from Boise State University
and founder and director of the Idaho Innocence
Project and has been involved in forensic
DNA testimony and cases like those of Amanda
Knox and Sarah Pierce, that I’m sure will
have heard of. He is a true Renaissance man. This is quite a bio– also
an award-winning playwright and a fellow of the National
Academy of Inventors. He holds patents
for power generation and a miniature pump using
magnetic shape memory alloys, whatever that is. This cannot help to be a
fascinating talk, so Greg, please welcome. [APPLAUSE] -First of all, happy
wrongful convictions day. Honestly, that’s what today is. It’s the first time the
International Innocence Network has decided to
declare this that day. And another coincidence is
that four years ago today, I was at my home in
Boise, actually just flying out
to CNN in Atlanta because we were going to hear
about Amanda Knox’s appeal. So she had lost at trial. She had been incarcerated
for four years in Italy. At that point, I’d been working
on the case for three years, I think. And we were going to hear
the judge would free her. And her sister just
posted on Facebook that tomorrow’s the
anniversary of that release. So it’s great to be here to
talk a little bit about DNA and especially something
that I do a lot of now, which is post-conviction testing, that
is, people who claim innocence and write to us. I work at the Idaho
Innocence Project. I volunteer as the
director over there. But I helped start
the one in Ireland, worked with the Georgia one and
really all over the world now. Because I’m the
only scientist who’s the director of one of
these organizations, they send me all the really
hard cases, hopeless ones that take a long time to solve. And enjoy that work quite a bit. So I’m going to talk
about some advances in DNA and I’m going to talk about
some problems with DNA, because I’m a bit
of a person who likes to hunt down problems. Last time I was at
Harvard– I think this was the last time– I was
here for an Innocence Network meeting. And the three
gentleman in the front here are three men whose cases
I worked on for the Georgia Innocence Project, just to
show you that it doesn’t all stay in the lab. And one of the people in this
picture actually was a student and helped free two of
them as his internship. Most people who
volunteer with us leave very disappointed,
because as I say, these cases can take
10 years and they want to stick around for six months. People always ask
about death row. And Charles Fain, from my state,
I did not work on his case but was spent 18
years on death row for the rape and
murder of a little girl nearby near where I
live and DNA got him out– mitochondrial DNA
got him out of prison. But if we speedily dispatched
the justice system, people do tire of
justice taking so long– Charles Fain is not one
of those people– he would have been killed. The average time on death
row is over nine years before someone is exonerated. So a lot of these people
that are being killed, of course the network
groups do not get involved with their cases. We don’t know how many innocent
people have been killed. So I want to show you
one of the great things that DNA can do and
one of the things that we don’t do so well. So I was talking with some
computer science people last night at supper and they
said, one of the things people do better than machines is
facial recognition, still. And that’s true up
until this point, though there’s been
a lot of improvement. So this is a standard lineup. This is actually a lineup that
John White here in the center was called for. He was working. They said, can you come in? There’s been a rape. An older woman who was 74 had
been raped in the neighborhood and beaten. And they said, look, we can
just get this over quickly. She’s going to go
through the lineup. He said sure, came in. They’re like, where do
you want to stand, John? He’s like, right in the middle. Let her get a good look. And she picked him. And so that was 1980. The student that I showed
you in the first slide worked on this case. And we actually got John White
out just a few years ago. And when we got him
out, we got a hit to the database of
the actual rapist or whoever left the semen
on this woman who was raped. And we were very excited. We got the name. And the student said,
I’ve seen that name. And I thought, well, a
lot of people write us. And some of those
people are guilty. So maybe this guy
wrote us from prison. He was in the system. That’s why his DNA
was in the system. He said no, I know
it’s from somewhere. So he went through all
the stacks of information that we had on this
case and found out that James Edward Parham,
by absolute coincidence, happened to be there
at the jail that day. They stuck him in a lineup. They’re like, you’re
just a filler. If she picks you,
don’t worry about it. You’re not a suspect. That’s how bad we are
at facial recognition, even though we’re better
than the computers. So DNA cleared this up. And so I want to say,
yay, DNA is great. And then I’ll talk about
some of the problems. So imagine that we can find
a gene that causes crime and we could just eliminate
that gene from the gene pool somehow. That’s kind of a
part of what I posit in the piece I wrote for the
New York Times that was titled by them, “Men: Who
Needs Them?” which got me hate mail from men’s groups. And I was like,
where were you when I was an angry divorced man? So anyway, these are
mostly guys who’ve lost custody battles
who were very upset. There was only one I
turned over to the police, but I got hundreds
of these letters. And I was just talking about
the Y chromosome and the fact that most of my clients in
prison for violent crimes are men. And so we have the gene. Well, we know where it is. It’s on the Y chromosome,
if there’s a gene. And should we do
anything about it? But mostly I was
extolling the virtues of the female half
of our species, which I consider the
only living half. I think the men– we only give
three picograms to that baby. She’s losing bone mass. And if you added up all the
mass of the male contribution to humanity from the beginning,
all these millions of years we’re talking about, it’s
less than a teaspoon. So anyway, I’d recommend
you take a look at that if you want. But the way I got
into forensics was I was working on the Y
chromosome in marsupials. So I titled this
“DNA is Universal: From Fruit Fly Eye Color”–
which I worked on at University of Connecticut, down
the road, for my Ph.D., as part of my Ph.D.–
“to Murder and Rape.” And “marsupial sex
determination,” I threw in there because
that’s what I did in between. I went to Australia– thank
you very much, taxpayers– on a National Science
Foundation grant, to study marsupial
sex determination and was part of the group
that I’ve found SRY, the sex-determining region,
our collaborators in London did– turned a female blastocyst
mouse into a male mouse. And that allowed me to give
talks all over the world. It was a lot of fun. Then I became a house husband
after that for a while to take care of our kids. So I learned a little bit about
what a lot of women go through. But because of the Y chromosome
work, I started getting calls, could we identify a stain as
male or female if it was spit? Now, this was 1990. And I though, well, yeah,
we could do that using SRY, or blood or anything else. And so I started
consulting a little bit in crime in the early ’90s. And just to show you how
universal all this is, the eye color of drosophila,
this is a kit developed, the IrisPlex system by Manfred
Kaiser and others in his group, that now we can look at a
stain, because of their work and others, and tell what eye
color, it is if it’s light blue and if it’s dark brown. Now, the hazel people you give
us a little bit of a problem. And to show you just how
sensitive all these tests are, it’s the single loci,
these single areas of the genome where just single
base changes can make changes in eye color that
are predictable. And so we used to go to
all the forensic meetings and they’d scan our eyes
and they’d take our DNA and they built this database. And apparently it’s very good. I have not seen
this used in court or in any crime solving yet. But it’s a pretty
robust kit and has been reviewed over and over. So my lab got involved with
some of the newer technology as well. My mother lived
in Paris and so I got to know some
gendarmes as young folk and then they went on and became
fairly important in the police departments. And they called me
about a case that had languished for 10
years where a woman was– she made a 911 call
and all you can hear is her screams and
then a couple of men. And then they found
her body burned, but some semen was recovered. They got a full hit. I mean, they got a full profile. Every bit of the DNA that
they wanted to measure, they measured, put it
into the French database, nope– no hits. So no criminal was
already in the database. And they called me
up and said, look, we’ve heard something
about familial testing. Nobody’s tried it in
continental Europe. They’ve done it in England. Do you think we
could do it here? We don’t have the software. And so I hooked them
up with some friends who had the software. It got all the way up to the
president’s level in France and the president
said, no, we’re not going to do familial
testing in France. And so I said, well,
there’s one other trick. And if you just set the
dial on the match stringency that the FBI software
has, the CODIS, you could just set it for
a half match, basically. Don’t find me a perfect match. Find me half of the chromosome
markers that are in common. Who would that be? Dad, son– right. And so that’s what they did. And I thought you probably
got 1,800,000 in the database. I had no idea. I’m not very good with the math. I didn’t how many hits
they were going to get. But I thought we’d get
at least a few dozen. Turned out we got just one hit. So this is just part
of what the crime was, the facts of the crime. They had a condom
with some semen recovered, cigarette butts,
nothing for 10 years. 1,800,000 convicted
offenders and arrestees, and we got one hit. We then confirmed that with a
male-only Y chromosome test, then got a sample from
the putative mother, found out all the missing alleles
were there, and found out that the person that it
matched as a potential would be a son of this guy. And he had a son about the
right age who had died, Gregory Wiart, and
he died in 2003. And so before they
exhumed the body– I’m guessing that they tapped
all the phones and found out when they started digging up
the body, who was making phone calls among his old friends. They got a voiceprint
or a voice expert to identify the voice
of one of the friends. And they think that was the
second voice on the 911 call. So that’s the way these
things are starting to evolve. And even if you were
not in the database, even if you’re not
in family tree, whatever, eventually
we’ll be able to find you. We’ll just reconstruct
the missing bits and for the purposes
of identification, we will probably be
able to find you. Right now, we need to
confirm it with actual tests. Something new that
we got involved with– I have a very
similar case in Idaho. Chris Tapp was a man who was
interrogated for 35 hours. I think– was it Dateline
who did the show? One of the big shows came
in and did the story. Anyway, and after
all those hours, they offered him a
deal if he would just say who the other guy was,
because the DNA didn’t match crisp. But they thought
maybe he knew it was. If he would just say
who the other guy was, they would give him
no time in prison. And so they’d been
interrogating him for 35 hours, telling him the
name of this guy, and so he gave them the name. They tested the guy
and it wasn’t a match. They came back. They said, Chris, it wasn’t him. Who else was it? So he mentioned everybody they
mentioned in this small town during the interrogation. They tested 54 people, I think. Chris was crying
at the end, saying, I’ll give you my mother
if you think it’s her. Who do you think it is? Anyway, so they revoked
his deal and charged him with murder and rape, even
though the DNA on the semen, on a dead girl’s body,
didn’t match him. And we’ve tested things
since then– pubic hairs, we’ve had them tested, that
were found on her face, et cetera, et cetera,
clothing that they tested– everything matches one guy not
in the Idaho criminal database. So working with the
police, we suggested, what about looking for the
last name of the semen? Treat the semen like an adopted
child looking for its daddy. And so they did that. I think they used methods that
maybe I wouldn’t have used. And it’s been a
bit controversial. But they did genealogy testing,
got the last name of someone, tested that person. It wasn’t him. And that shows you the
problem we then have. We have to go to the
papers, the paperwork and find out who else is
related to that Y chromosome. But that’s the kind of thing
that’s going to happen. And I think it was mentioned if
everybody was in the database eventually, how much easier
all this stuff would be. This is kind of
the results you get when you do that type
of search of publicly available databases. But sometimes you get people who
don’t want their name listed. And so you won’t
get a last name. And a court order was
used in that case, in the case I talked about. Some basic DNA– let’s look
at how it’s done just really quickly. Here’s this cell. Hopefully, you’ve
seen one before. That’s the nucleus. This is a gigantic mitochondria. They’re not that big. There could be 1,000
of them in the cell. We get them only from Mommy. Like I said, Mommy
gives us everything. Daddy gave us a
little bit of sperm and he’s actually
not even there when the nucleus enters the egg. He’s already gone. It happens a day later or so. But anyway, the mitochondria
come only from Mom. There’s thousands of them, or
thousands sometimes of them. So we can even get mitochondrial
DNA from a broken hair. Mitochondrial DNA
was famously used to identify the unknown
in the Tomb of Unknown from the Vietnam War. So there is no unknown
from the Vietnam War. And the family had to
fight to get that done. The military resisted. But anyway, that tomb is now
not only unknown, it is empty. We get our DNA from swabs. I carry them in my
bag wherever I go. And basically it’s a
Q-tip type of thing. And the cells look like this
unless they’re sperm cells. Sperm cells are really tough. They have a
disulfide head and so we can separate them
from all the other cells. It’s the only type of
cell we can separate. And so that’s how
in a sexual assault we can get what
we call the male, profile all the
male contributors. And we don’t care about
the three billion bases. We only want to know
the size of 13 areas– how big a piece you’ve got in
the sperm, how big piece you’ve got in the egg. So for several of the
chromosomes, we do that. If we have 13 of those
areas, just the size from Mom, the size for
Dad, and their number is, like, five, seven–
very, very simple numbers. We know those sizes. We can find you. So there’s the cell. We magnify it a few hundred
times, again, a few hundred more times. If we could see the
bases, it’d be like this. And we’re just going
to amplify this with something called
polymerase chain reaction that won the Nobel Prize
for Kary Mullis. Here are the 13 areas that
the FBI currently requires. They’re going to be
expanding soon to over 20. But these are the chromosomes. And chromosomes are
nice because they were discovered before
anyone really knew how they were working very well. So the big one is called
1, the less bigger one 2. There’s one actually
kind of out of order. And then you have
the X and Y here. Is this a man or a woman? It’s a man, right? Y chromosome. So if we have a Y
chromosome for humans, we’re generally a man, unless
you have a mutation in SRY or a couple of other genes. All right, here’s the data. Ready? I want to teach you how
to read one of these. This is called an
electropherogram. Can you say electropherogram? electropherogram. Great– you’re on your
way to testifying. So now those areas, those 13
areas, this is one of them right here. This is from chromosome 3. And you see the
size is 16 and 17, so Mom and Dad had
two different sizes that they gave this person. So this is another
location on a chromosome. Mom and Dad gave the
same size there, 15. And that’s why the
peak is higher. It means there’s more DNA. Here’s another one, size 21, 22. When I find then the
frequency of the 16 allele in the population, I can
multiply it by the 17 allele and I can find out how often
I would expect this profile to come up in the population. That’s the statistic
I would use. So this is a single contributor,
no more than two peaks at any one locus, location. Now, let’s talk about
some of the problems. I’m very concerned about
the problems in DNA, because I think we’re
having a large number of wrongful prosecutions
with DNA mixtures, where there are
mixtures, or whether it’s very small amounts of DNA
like in Amanda Knox’s case. And she was convicted based
on a very small amount of DNA that they found on
a knife blade that was consistent with the victim. And Amanda’s DNA
was on the handle. We said that that DNA
on the knife blade was spurious, probably
a contaminant. It was never repeated. It was below the level
of DNA the FBI would even look at at the time, or any
of the labs I worked with. And that lab hadn’t
been– the Italian lab hadn’t been accredited
yet, certainly not to work with that low template DNA. So just because
there’s DNA doesn’t mean that it’s good science. Report language can be very
weird and be very biased. For example, you
can call something a profile when there are
several people in it. So you can say the profile
matches the client. But there’s actually several
people in the profile. And oftentimes, that
lack of a plural confuses members of the jury. And then we can talk about
statistics and DNA transfer, if there’s time. I’ll get through as
much of this as I can. Peter Gill, one of the
great experts in our field, has said, if you show 10
colleagues a DNA mixture, you’ll probably end up
with 10 different answers. You won’t hear any
experts say that on the stand who’s coming
in for the state, though. They’ll just say
this was a match. Here’s my calculation. But all of us in the field
know that this is a problem. We didn’t know how big a
problem until [INAUDIBLE] and myself did a study and
now John Butler at NIST has followed up. And I think they’re
very telling, and I’ll show you what they are. So this is what a
mixture looks like. In any case, there’s
lots of peaks here. And there’s three peaks
here, so it’s obviously more than one person. So what happened was with
that data from that slide, we took that data– it’s
an actual case where I have– there’s
a man in prison, Kerry Robinson in Georgia. I believe he’s innocent. He was convicted of being in a
gang rape, a guy whose DNA was perfectly in that sample
from that poor girl and who admitted to shooting
people and to raping her, then was offered a deal. Tell us who the other
guy was, we’ll knock 8, 10 years off your
sentence, whatever it was. So he said, Kerry, and he
gave a wrong last name, and then, like, Kerry Robinson. And so they went and
arrested Kerry Robinson, a guy he didn’t like. And Kerry’s like,
I didn’t do it. All they have is evidence
for Kerry Robinson. The woman who was raped
picked the other guy out, didn’t pick Kerry out. All they had for
evidence was the word of this guy who
was making the deal and a corroboration of the
DNA scientist who said, I can’t exclude Kerry
Robinson from that mixture. I think he’s excluded. Eric Carita, who was at
the Connecticut Crime Lab, thought he was excluded. And Fox News in Atlanta did a
story on it and I said, look, this is a statistics problem. But I think we can explain it. I said, swab four people. Here are swabs. Four of your
reporters, swab them. They swabbed them
and I did the DNA. All of them were included,
according to the GBI. So we took that data and
we gave it to 17 scientists at a accredited North American
crime lab and asked them, is this profile
included, excluded, or you cannot conclude,
is inconclusive? We gave them no
contextual information. And most importantly, we
didn’t allow them to talk. Like, if I allowed
you to talk, you might all decide I’m
giving a good talk, right? But there’s 15 of you in
here who think I’m terrible. We do this
groupthink– maybe 10. We do this groupthink and
we come to conclusions. Oftentimes, you do a
straw poll of a jury, they find the client innocent. By the end of the day–
that’s their first straw poll– by the end of
day they can convict. We process things
socially different than we do when we’re alone. And so we did the study
and out of those 17, only one scientist
at that lab agreed with the Georgia Bureau
of Investigation– only one of the 17. And we published it in a
peer-reviewed journal, Science and Justice. And it got a lot of press. So The Economist
picked up on it. New Scientist picked up on it. One out of 17–
I’ll just skip that. So then I had a lawyer write
the head of the Georgia Bureau of Investigations
DNA, like, what rules are you
actually applying? Because it seemed like even if
you had one allele in common with all of those
markers, just one– and I have one in just
about every case I’m with, they would say you
cannot be excluded. And in fact, what the
head– you can’t read this– but what George Herrin
said was even if there are no alleles in the
evidentiary sample consistent with or matching
those observed in the subject’s known DNA profile, no
alleles with the evidence, this does not constitute
a conclusive exclusion of the subject, since
the absence of evidence is not the equivalent of
the evidence of absence. What this means in
practical terms is, anybody who wants to reduce
their sentence could point to any of you in the room and if
you had zero DNA in common with the evidence, that would
be a corroborating fact, because you cannot be excluded. That’s all it takes. So while I can understand how
George might have said this– like, what he’s trying
to say is, I can’t say– he should have said
it’s indeterminate, it’s inconclusive– not,
it’s cannot exclude, which in DNA language–
he should know– in DNA language means
you’re included. So there were no stats
used in the report that convicted Kerry Robinson. Thank goodness 2010,
the scientific working group on DNA analysis
and methodology says, you have to
have statistics. Hurray, we all say. Now there’s going to be science. So John Butler at the
National Institute of Standards of Technology does
something like what we did. He starts a study in
2005, actually before we did, and just
gives the same data to a bunch of volunteer crime
labs all over the country. They pick out the right people. And then he asked
them for a stack. Give me a matched statistic. Now, when you go to court,
you have damage to your bumper and you have to have
two receipts generally, one for $30, one maybe
for $45– the judge chooses the one in the middle. With DNA, he gave
the same DNA and what did the stats look like? These are the stats
for one of the samples. And the conclusion was some of
the labs had it at 10 to the 5 and some had it at 10 to 15. There were 10
orders of magnitude. That’s a difference between
a $30 damage to your bumper and a $300 billion damage. So you should get
the $150 billion, minus, give or take, $30. So that was 2005. 2013, we’ve published our paper. I thought, finally they’re
going to do something about these mixtures. I’ve been talking with
John Butler at NIST and asked him to redo the
study– please, please, please. He always intended to. And please, please, please,
publish your last study. He never did. I’m hoping he’ll
publish this one. And they had five
different cases in 2013. They sent the same data out
to all these volunteer labs. So four person mixture,
four people from John’s lab add DNA to this thing and
they pretend it’s a ski mask, send the data out
to all these labs. They also send the labs, two
of the lab people’s profile who were actually in the mixture
and one person who wasn’t. And they say, who’s in,
who’s out, give us a stat. 108 labs– C is the person
who is not in there. Out of 108 labs, only
seven excluded C. You could go to prison for the
rest your life on a mixture. And this is the kind
of nonsense that’s actually going on in this
country and all over the world. This is not science the way
it’s supposed to be done. And DNA got a pass when the
National Academy of Sciences looked at all of the
forensic sciences. They didn’t like bite marks. They criticized
fingerprints because they don’t have a database. They love DNA. And I was just very upset
that day when that came out. So out of those
labs, 76 of the 108 included the person
who wasn’t in it and provided a match statistic. The match statistics
weren’t like, 1 in 10, could be anybody. It was one in nine for some
labs– could be anybody– to one in 344,000 that I
guarantee you would convict. Where else does this happen? This is Taiwan, Taiwan
Association of Innocence, a new organization, came to
one of the national meetings in 2013. They were telling me
about this guy who was arrested because there were
two escorts who had been raped in a warehouse and this guy had
an interest in the warehouse. There were three suspects. Two of them were convicted
in 2010, so a year later. And then they start
prosecuting Mr. Chen also, who says he was innocent,
never had consensual sex. The other guy said they had
consensual sex with the women. So here’s the Y chromosome
profile from the underwear. This is the victim’s underwear. This is a typical
way we work a case. These are just locations
on the Y chromosome and what size DNA
each person has. So that’s Mr. Chen, Suspect 1,
Suspect 2 who were convicted. And as you see, at the first
location, Chen is a 15. What’s the underwear? 15, so it’s a match. Any time I say
match, my students have to say statistic–
so match, statistic. Actually we can’t
calculate it for this. But that’s probably, let’s
say, 1 in 10 for the purpose I’m going to put here. A 12 allele, they’re all 12s. Well, there’s 12
in the underwear. We go through– guess what? Poor Mr. Chen, he’s
in the underwear. It’s a match. That’s now statistically
pretty robust. We have all of those
parts of the Y chromosome. In fact, that’s the entire kit. We can’t go any further. What do we say? I said to the
people, why don’t we see if we can do more
locations on the Y chromosome. Maybe he’ll be excluded later. So they go to court
and more of this date. So he’s guilty in 2012. His appeal’s denied, 2013,
we take the case, 2013. And we go to court and the
Taiwan lab worker said, we have a new test. It’s got more loci. It’s got six new
locations in addition to the other ones that
were already there. And there’s Mr. Chen. He’s a 17. He’s excluded at two
of the locations. There’s Mr. Chen. And finally I got
to meet Mr. Chen, because he could travel
just a few months ago. So here’s again a case
where you have a DNA match. It looks good. It looked good even to me. The full kit was used. But it’s a mixture. And these mixtures in
reality are very difficult and they are confounding juries. Now, I told you about
the mitochondrion. Mitochondria are good
for us because if there’s a fraction of a hair
left without a root, we can use mitochondrial DNA. We do a lot of mitochondrial
DNA on the Basque. I study the Basque
in my lab as well and we have some
projects on them. But anyway, the
mitochondrial DNA is something you get
from your mother. We can do lineage
stuff, and it can be used when there’s
just a fragment of hair or the body’s very
decayed or burned. And Donald Denman– this is
from the Albuquerque journal. He went missing. He was 49 years old. His family struggled. They never had the body,
never found anything. Finally, some bones are
found near the house. And so they identify
those bones. This is February of
2008 and donations could be sent to the Salvation
Army for Donny Denman. The FBI confirms the DNA by
mitochondria of those bones. Since it’s from the mother, they
get one of the mom’s relatives, one of her sisters
in another state. They finally confirm it. Now we’re in March
of 2008, same year. It’s tough for his
brother to look at the photograph
of his brother. As he stared as the image,
he couldn’t help but wonder, would his brother have
turned his life around if he had lived? This is now May 3, same year. I said match. You didn’t say statistics. So with mitochondria, it’s
generally one in 2,500, one is 5,000,
something like that. And so it was a coincidental
match to somebody else’s bones. A friend of Donny
Denman, who was repairing his car in his garage,
putting down newspaper, sees the obituary,
calls Donny and says, Donny, they buried you. And he’s like, you’re kidding. He’s like, well, who
spoke at my funeral? Like, your pastor. So he called the pastor. And like he says,
I’m very much alive. It’s a trip, isn’t it? Once again, good
DNA, a true match, a statistic that
is not an identity. It is a good statistic. 1 in 2000 sounds good. 1 in 5000 sounds good. So now we’ll talk
about sensitivity. Can we become too sensitive? Here is a slide of some sperm. When I was first doing DNA in
the ’80s, we’d need about 500 of these slides to give a
really good DNA profile. We were using a technique where
we couldn’t amplify DNA yet because it hadn’t been invented. Now, single cell. So I want you to consider this. As our instruments and our
kits have gotten, like, a millionfold better,
yet most police work is still collecting
evidence the same way. Let’s think about what
a problem might be. So I try to boils down to
people’s practical experience. How many of you swim
in a public pool, have swum in a public pool? Great. So the problem is this. What if one person,
man or woman, goes into that public pool
with an ejaculate on them or in them? And I got news for you. As you’re swimming, that
water is doing more than going over your body. It’s cleaning you out a bit. And if one person goes in, 300
million sperm cells from one ejaculate in that
public pool, how much water do you have
to swim through before you get a sperm cell? Well, there’s 300,000
gallons in an Olympic pool. It’s pretty easy. There’s 1,000 sperm
cells per gallon for every ejaculate
evenly dispersed. Now, if you happen to get
a rich area of ejaculation, swim with your mouth closed. So in 1982, I’d need
1,000 of those gallons to do a DNA profile. Today, I need about a teaspoon. So how does that work
into forensic casework? Well, Meredith Kercher
was the young lady who was killed in Italy,
the student who was killed that Amanda Knox
served time for and was eventually exonerated for. That’s Amanda and Rafael, her
co-conspirator who supposedly were involved in the murder. They were suspects
before any of the DNA from the room or the
fingerprint evidence was done. And so it was a gut feeling
from the prosecutor. And, can you have gut
feelings in science? Yeah, they’re called
hypotheses, right? But the difference between a
theologian– some of my friends who went into
theology– and me, who went into sciences, a theologian
will die for their ideas. And that is the measure
of their dedication. And a scientist’s
measure of dedication is you let your ideas die. You put that baby
that you’ve– I mean, I’ve tried to cure
cancer for a long time. You put it up for
a test every time. And everything you thought
was going right, you abandon. There was a lot
of press on this. They spray a tan on you at CNN. That was before the beard. So, only two people’s
DNA– when they finally do the DNA and the
fingerprints, only two people are in the murder room,
the victim and Rudy Guede, who they try and
they put in prison and they still persist
with Amanda and Rafael, based on that gut feeling. Finally, we had the ruling
from the Supreme Court just a few weeks ago
that said, no, they got it completely wrong. She’s completely innocent. It was ridiculous. She should have been
released immediately. He’s serving time. This is the bra clasp that
convicted Rafael, Amanda’s boyfriend of a couple weeks. They picked it up 46
days after the murder, after the teams had been
through a couple times. They said they forgot it. I could show you
the collection film. They actually have it. They’re passing it around. They’re touching it, passing
it from one guy to another. And then they put it
back on the ground, because they forgot
to photograph it. So here they are, totally
in the bunny suits and they’re putting
evidence on the ground. And they said they’re
changing their gloves and the gloves are
clean and we had evidence their
gloves aren’t clean, and we showed all that video. That’s the knife that
supposedly Amanda used. And that’s where DNA consistent
with the victim was found. This is the electropherogram
of that DNA from the victim. What is important here
is that the y-axis, which the FBI at that time wouldn’t
look at anything below 200, there’s only one
peak above 100 here. They were never
able to repeat this. It was an invalid procedure
and that’s why she’s free. We repeated this in my lab
as we were studying it. We collected soda cans from
the dean’s office at lunch. And then I told my people,
save money on the gloves, just like they apparently did. Don’t change your gloves. Change in every other
piece of evidence. Collect a can, go
in another room, and unpack a knife
from the Dollar Store and put that in a
separate evidence bag. Keep everything separate,
then change your gloves. Do it for five people. We did it for five
people and we found one person’s DNA
transferred to the knife– one out of the five. I’m going to skip through this
just because we’re at the end. So, happy ending. There we go. Amanda, that’s a few
weeks after she got out. We had Thanksgiving
together at her house. And then both of our
families went rafting for three days in Idaho. I’ll show you one more
example of how this can go. And somebody’s going
to have to shoot me to get me off the stage. I mean a photograph. My mom lived in Europe, so I
followed this story for years. For 15 years, this woman
was committing crimes all over Europe. Rarely was there a witness. There’s a murder. There are Romany
chiefs, feuding Gypsies who use guns and her DNA
is found on the guns. She robs. She injects herself with heroin. They found syringes
with her DNA. Kills with almost
professional precision. As far as German detectives are
concerned, she has no identity. They test 800 women
in Heidelberg. They do DNA test on
her– nothing, no match. Bruno Bosch says in
2008 he’s traveled 60,000 kilometers questioning
witnesses– nothing. They did 800 women,
as I told you. And in the end, it was a woman
who worked in the Bavarian swab factory. She was leaving a little
something on her swabs. That’s the Phantom of Heilbronn. Why wasn’t this detected? Because we don’t track our
errors in forensic science. And I’ll show you my
lab’s answer to that. We study the smallest sequences
that don’t exist in nature. We call them nullimers. For DNA, they were
recently 13 bases long. For proteins, they were
five amino acids long. And we call them nullimers. And we’ve made cancer-killing
agents out of the proteins. But out of the DNA,
what we’ve done is we’ve made a marker
that amplifies better than any of the human markers
and it looks like this on an electropherogram. Those are human markers
that have been diluted 100,000 times after PCR. And then we diluted
it a million times and tried to PCR up everything,
all the human markers and our nullimers,
put it on a knife, and you can still see that. And so that means when you
give a swab to the police, at least one thing we can avoid
is they make a billion copies. That’s what I do in my lab. You give me DNA, I
make a billion copies– first thing I do. Each of those copies
is indistinguishable from the original in PCR. It’s not like a
traditional fingerprint. The Supreme Court got that wrong
when they said that it’s just like a traditional fingerprint. We can take them from arrestees. It’s not. I make a billion copies and one
of those copies in an aerosol, on a tube, on a glove,
is enough to convict you. And that’s why we’ve
invented the nullimer marker. So I’ll skip all this. If you’re interested more in
hearing– because we’re out of time, right? You can read my book or contact
me at [email protected] Thanks. [APPLAUSE] -We’ve got time for about
four or five questions. If you guys want to come–
is the microphone out? I can’t quite see. We’ve got a microphone. I guess we can– in
the middle there. Come on up, say who you
are, and if you can just keep it to a question,
that would be awesome. And maybe while we’re
waiting for folks to come up, I’ll just ask you. If you were to recommend one
thing to the police, one thing to the labs, one thing
to analysts to improve this system, what would it be? -What I tell my students in
lab is keep your arsenic out of the spice cabinet. And for God’s sakes, if
you keep them together, don’t keep your arsenic
in an Old Spice bottle. So what I’d say is
these are the– I’m going to do it by demo. We work in very small
volumes in lab now and these are the racks. And you can’t possibly
read the writing on the little tubes
in here, so it’s all about the order they’re in. I skipped a case. I have two cases where they’re
just swapped tubes that that’s what put people in prison. And Las Vegas just apologized
to a guy they did that for. So I would say, keep
suspect things separate. They should go to
a separate lab. There should be one national
lab or maybe a couple national labs that
just do known samples. And then the rest of the
crime labs do their thing. But they’ve got to keep
them separate all the time. I think there’s a lot
a mixup being done. And the tragic thing about
those is there’s no way to tell. There’s nothing I can do for
somebody if that’s happened. -Questions. -Hi. Great talk. Given that we know that
there are trace amounts of genomic DNA
all over the place and it’s quite hard to amplify
STRs and stuff like that from trace amounts,
what do you think the probability is that the
forensics community will turn away from STRs and mitochondria
and to, like, low coverage genome sequencing? -Well, it’s funny
you should ask. So the FSI Genetics
that I just picked up in the mail on my way
over is “New Trends in Forensics Genetics.” And it’s all about
massively parallel sequencing, even things
like microbial genetics. Do your microbes identify you? Probably. I want to do a study
actually about people who take antibiotics
and then get divorced. It is because after years, we
develop the same microbiota. And if one partner’s taking
antibiotics and going traveling, you change
your fingerprint. Does it change your behavior? Does it change your interests? Do you find a different
smell pleasant? So they’re looking at microbes. That’s all free stuff. That’s my idea. You can take it and run with it. But they’re looking at
other ways of mostly just massively parallel sequencing. You can now– and with
mixtures, I can only tell three people apart. They can look at 1,000 people
and say if one person is included or not. So it’s pretty impressive,
but nobody’s used it in court. Yeah? -Hi. I’m Denise McWilliams. I’m with the New England
Innocence Project, and thank you for your
help to us over the years. -Yeah, you’re welcome. -I recently was at a
conference in New York where I watched the
scientist and attorneys for the prosecutor
and the defense literally scream at each other
about the use of low copy DNA. -That was in New York? -There was in New York. -Yeah, of course. -Just curious what
your thoughts are about the use of low copy DNA. -Well, low copy means
two different things. The original meaning
of that meant we did extra rounds of PCR. And in that case, you actually
see what we call drop-in. And all the drop-in
is, is DNA that’s just kind of floating
around or in the glove when you buy– on the
glove or in the tube that you don’t see with these
normal 28 rounds or 30 rounds. But when you do the extra
rounds, you get drop-in. But low template
DNA, which probably is something they
were arguing about, is when you have very low levels
like in Amanda Knox’s case, you’ll get different results
in some of those cases. And so in New York
at the OCME, they were saying, oh, if you get the
result two out of three times, you can report it. And when I did a
case in London, they said, well, it has
to be repeatable. And I thought, that makes sense. A repeatable result is reliable. And then I looked
at their records. They did, like, 10 tests. Two of them were
what they wanted. They picked the two. They called it repeatable. So I think there’s
a lot of problems with some of those techniques. But frankly, that
bar keeps moving. As we get down to
single cell resolution, it won’t be as much
of a question anymore. So we keep lowering the bar
of what low template is. So, yes, it should be used. Yes, there are problems. -Hi, I’m Faye Flam. I’m a journalist with
MIT Technology Review and I was intrigued by these
cases where there are mixtures and the statistics looked
good, where there was like a 1 in 2000 chance it was wrong. Is the problem there
that the person was the 1 in 2000 or the problem
that people are screwing up the numbers, that they’re
making statistical errors? -The problem is that there
are several good methods to do statistics
and they produce extremely variable results. So when I say good, they’re
based on mathematics. All the statisticians agree
they’re good, robust methods. But you get 10 to
the 10 differences. And then you get labs
that will exclude. Most labs are still
applying rules of thumb. If there’s only one
allele, we won’t call it. Utah told me there
had to be at least two that weren’t found
to exclude somebody. They all have these rules of
thumb that they’re applying. So the statistics are helpful. But also, I think our idea
about these statistics is wrong. If I say 1 in 5,000 in
a Boston crime, that means there’s 1,000 people
in Boston who might– who are I match, a
definite match, right? And so if you think of it
in terms of epidemiology or in terms of medical
testing, that’s a false positive
rate of 999 to 1. So for every one person in a
1 in 5,000 match in Boston, for every right one,
there are 999 wrong ones. And it doesn’t mean
we don’t use that, but you have to start
giving a better perspective. Because jurors hear 1 in 1,000,
1 in 10,000, 1 in a million, and they think it’s over. They don’t hear
numbers like that. They hear 50-50. It’s 20% chance. When they hear those kind
of overwhelming numbers, they generally convict. -Thank you. [MUSIC PLAYING] -Thanks. And I’ll be around if
you have questions.

Add a Comment

Your email address will not be published. Required fields are marked *