Gene Machines: Navigating a World Enthralled by DNA

CAS Author

11/08/2019 Added

27 Plays

Description

Biologist Colin Meiklejohn gives the third CAS Inquire lecture, connecting genetics to the 2019-2020 theme "Rise of the Machines." More information is available at https://cas.unl.edu/cas-inquire.

Searchable Transcript

Search:

[00:00:05.010]Good evening, I'm Taylor Livingston, and I'm the Director
[00:00:07.720]of the College of Arts and Sciences Inquire program.
[00:00:10.750]Thank you all for coming this evening
[00:00:12.470]to the fall semester's final installment
[00:00:15.290]of this year's Inquire Series lecture,
[00:00:17.987]"Rise of the Machine."
[00:00:19.840]The Inquire program is structured around these lectures,
[00:00:22.630]allowing students, faculty, staff,
[00:00:24.950]and the wider public the opportunity to investigate
[00:00:28.380]how we as a society engage with technology.
[00:00:31.730]Not just its benefits or negative aspects,
[00:00:35.010]but the nuances and effects of our cyborg entanglements.
[00:00:38.770]It creates an opportunity to learn
[00:00:40.770]about the fascinating research of our faculty members
[00:00:44.160]in the College of Arts and Sciences,
[00:00:46.360]and enables students to see various
[00:00:49.923]disciplinary approaches to the study of a topic,
[00:00:52.810]as well as the necessity of a multi
[00:00:55.060]and interdisciplinary insight
[00:00:57.210]to truly understand social issues.
[00:01:00.610]For a distinctive set of intellectually curious
[00:01:03.310]and inquiring students, the Inquire scholars,
[00:01:06.140]if you will, please stand.
[00:01:09.530]The program offers a chance to foster a meaningful
[00:01:12.420]academic engagement with the topic through discussions
[00:01:15.210]with the lecture faculty, and a year long research project.
[00:01:19.620]They are aided along the way by the peer leaders,
[00:01:22.040]if you could please stand.
[00:01:26.714]If you could please join me in thanking the students
[00:01:28.840]and their participation in the program.
[00:01:31.126](audience applauding)
[00:01:36.020]Tonight I have the distinct pleasure of introducing
[00:01:39.030]Dr. Colin Meiklejohn, Assistant Professor
[00:01:41.601]in the School of Biological Sciences.
[00:01:44.570]Dr. Meiklejohn's work focuses on evolutionary genetics,
[00:01:48.050]particularly how new species emerge,
[00:01:50.590]changes in gene expression, and the co-evolution
[00:01:53.730]of nuclear and mitochondrial genomes using the fruit fly.
[00:01:57.740]Tonight his talk, "Gene Machines:
[00:01:59.527]"Navigating a world enthralled by DNA"
[00:02:02.470]will help us make sense of what is fact
[00:02:05.150]and what is science fiction
[00:02:06.760]when it comes to direct to consumer genetic testing
[00:02:09.830]such as 23andMe and AncestryDNA.
[00:02:13.240]Please join me in welcoming Dr. Meiklejohn.
[00:02:16.254](audience applauding)
[00:02:38.790]All right.
[00:02:40.270]Thank you very much for the opportunity to talk to you,
[00:02:43.170]I'm very excited to talk about DNA
[00:02:46.150]and the machines that we use to study it.
[00:02:52.340]So, 67 years ago,
[00:02:55.220]the structure of DNA was elucidated
[00:02:58.020]by Franklin and Watson and Crick,
[00:03:02.150]and provided us with this iconic structure,
[00:03:05.040]now the double helix, a single molecule
[00:03:08.940]made up of four letters, A, G, C, and T,
[00:03:12.630]that is the underpinning for all of life on Earth.
[00:03:17.560]And since then, science has not stood still,
[00:03:23.150]and we have made really amazing strides
[00:03:27.110]in our ability to understand DNA, to identify it,
[00:03:30.650]to sequence it, and to try and figure out what it does.
[00:03:33.770]And so these are the machines that sequence DNA.
[00:03:39.790]In the last 15 years or so,
[00:03:45.510]the decrease in the cost of sequencing of DNA
[00:03:48.230]has really put Moore's law to shame.
[00:03:53.070]To give a example, in 2018, I did an experiment,
[00:03:58.900]my lab where we sequence DNA,
[00:04:01.690]and we got 92 billion base pairs of DNA
[00:04:07.020]for about $9,000.
[00:04:09.420]And so that is more than 100,000 bases,
[00:04:12.820]A's, T's, G's, and C's for every penny.
[00:04:16.090]And this low cost of sequencing
[00:04:19.600]has been driven mainly by technological advancements,
[00:04:23.270]optics, microfabrication, chemistry.
[00:04:27.230]So for example, there are now multiple methods
[00:04:31.070]that are common for sequencing DNA at high throughput.
[00:04:38.020]One of them involves, for example,
[00:04:40.020]fabrication of these tiny little chambers
[00:04:43.550]which are less than 100 nanometers wide,
[00:04:45.460]which is about 700 times thinner than a human hair.
[00:04:49.260]At the bottom of this chamber there is a single molecule
[00:04:52.410]of DNA polymerase, the molecule that copies DNA,
[00:04:55.320]and as a molecule is fed through,
[00:04:57.940]it releases fluorescently labeled A's, T's, G's, and C's
[00:05:02.410]and this is read off by a laser
[00:05:04.930]that then records a movie of this happening in real time.
[00:05:09.350]And so based on
[00:05:13.900]just sort of a ballpark back of the envelope
[00:05:17.000]basement number, we probably are generating
[00:05:22.320]over 9 quadrillion bases of DNA per day on this planet,
[00:05:27.800]and that's been going on for years now,
[00:05:29.750]and so we are swimming in DNA sequence information.
[00:05:34.730]And I think the question is,
[00:05:36.410]what have we learned from all of this DNA?
[00:05:39.440]And where are we going to go as a society
[00:05:43.410]with all of this information and all of the promise
[00:05:47.060]and all the pitfalls that it holds?
[00:05:49.470]So, I wanna start by talking about
[00:05:53.710]how we use DNA information to make inferences,
[00:05:58.060]and throughout I'm gonna lean a lot
[00:06:00.210]on some really neat figures and concepts
[00:06:02.650]that were developed by Graham Coop,
[00:06:03.910]who is a population geneticist
[00:06:06.660]at the University of California Davis.
[00:06:09.260]So we all have, whatever your family situation,
[00:06:13.450]we all have two biological parents.
[00:06:15.750]We have a mother who contributed slightly more
[00:06:18.170]than half of the DNA that you have,
[00:06:19.850]and a father who contributed slightly less
[00:06:22.270]than half of the DNA you carry,
[00:06:24.000]and we can simply represent this,
[00:06:26.930]here's your genome, half from Mom, half from Dad.
[00:06:31.010]And then, of course, each of your parents
[00:06:33.700]had themselves two parents,
[00:06:35.770]so our genome can be thought of as being inherited from
[00:06:39.410]four grandparents, a paternal grandfather,
[00:06:41.553]a paternal grandmother, a maternal grandfather,
[00:06:43.383]a maternal grandmother, and so on
[00:06:45.590]back through the generations.
[00:06:47.450]So in general, we have two to the K ancestors
[00:06:51.190]K generations ago, and on average,
[00:06:55.320]we inherit one over to the K of our genome
[00:06:59.210]from each ancestor, so eight generations ago
[00:07:01.420]we have 256 ancestors, and on average,
[00:07:04.760]we inherited one out of 256 of our genome
[00:07:07.780]or somewhat over a million base pairs
[00:07:10.720]from each of those ancestors.
[00:07:13.900]But this average is somewhat misleading,
[00:07:16.730]and that's because in the actual process of transmitting DNA
[00:07:20.940]from parents to offspring,
[00:07:22.670]there are a number of steps that are highly stochastic.
[00:07:26.130]And that is what I'm illustrating here.
[00:07:29.080]One of those steps is that we do not transmit
[00:07:32.990]our chromosomes intact to our offspring,
[00:07:36.100]we do not inherit our chromosomes from our parents intact,
[00:07:39.700]but rather the chromosomes are shuffled every generation.
[00:07:42.550]And we call this recombination,
[00:07:44.600]and it involves physical breakages between the chromosomes
[00:07:47.330]and then stitching them together
[00:07:49.300]to make a chimera that is a hybrid molecule
[00:07:51.750]between the two parental copies.
[00:07:53.570]So what's shown here is one complement
[00:07:56.870]of the 23 chromosomes that we each have,
[00:07:59.760]so all of us have 46 chromosomes in ourselves.
[00:08:02.840]23 from Mom, 23 from Dad.
[00:08:05.390]And what's shown here is a hypothetical example
[00:08:08.250]of one compliment, let's say in a daughter,
[00:08:11.110]and these are, what's labeled here are the colors,
[00:08:15.470]are the segments that she inherited through her mother
[00:08:19.510]from her maternal grandfather and her maternal grandmother.
[00:08:22.970]And so you can see, for example here,
[00:08:25.430]these three arrows indicate these recombination events
[00:08:29.930]where she got the tip of one chromosome
[00:08:33.380]from her maternal grandmother and then a bit
[00:08:35.110]from maternal grandfather, grandmother, and so on.
[00:08:38.290]You can see that some chromosomes are transmitted
[00:08:41.080]unrecombined, so chromosome 11,
[00:08:43.900]this hypothetical daughter only got
[00:08:46.760]the chromosome 11 from her maternal grandfather
[00:08:49.160]and no DNA on chromosome 11 from her maternal grandmother.
[00:08:55.360]And this is only half of the situation,
[00:08:58.330]there's another complement that's coming down
[00:09:00.150]to the father that would similarly be scrambled.
[00:09:02.630]So that's one way that this process is more stochastic.
[00:09:08.500]And we can see this,
[00:09:10.020]here's another way of representing this.
[00:09:12.270]So here are 22 pairs of the autosome,
[00:09:16.440]so those are the chromosomes that are not
[00:09:18.280]the X and the Y that determine our sex.
[00:09:21.260]And you can imagine, looking at,
[00:09:23.560]this is my chromosome one
[00:09:25.920]painted on to my dad's chromosomes.
[00:09:28.440]And so what this kind of figure is showing,
[00:09:32.460]for example, chromosome four here,
[00:09:34.440]I inherited, half of it was from my dad's mom,
[00:09:38.310]and half of it was from my dad's father.
[00:09:41.510]And so that is evidence that there was a recombination event
[00:09:43.850]right in the middle here of chromosome four,
[00:09:45.970]in this example, you can see that for most
[00:09:47.940]of the smaller chromosomes, there in fact
[00:09:50.020]in this simulated example, there was no crossover,
[00:09:53.220]and so only one or the other chromosome was inherited.
[00:09:56.510]And so, if I were the only progeny,
[00:09:59.520]the only member of the lineage,
[00:10:01.490]then the alternate chromosome 14 would be lost.
[00:10:06.460]So this leads to how we can use this kind of framework
[00:10:11.540]to think about similarity between relatives
[00:10:14.320]and putting similarities between siblings.
[00:10:16.620]So here is, again, this picture of my chromosomes
[00:10:19.580]painted on to my dad's genome.
[00:10:21.630]This might be my brother's chromosomes
[00:10:23.460]painted on to our dad's genome.
[00:10:25.370]And then what you can see here is the overlap
[00:10:27.790]between my genome and my brothers
[00:10:30.850]with respect to the pieces that we inherited
[00:10:32.550]from our father.
[00:10:34.020]So, dark purple here
[00:10:37.380]indicates that we inherited the same chromosome.
[00:10:39.630]So for example, chromosomes 18 and 19,
[00:10:42.180]both in this case, myself and my brother
[00:10:45.550]inherited the same chromosome
[00:10:47.250]whether it was from my dad's dad or my dad's mom.
[00:10:50.680]Whereas for, where is it, chromosome 17,
[00:10:55.330]we inherited the alternate chromosome,
[00:10:57.350]so my brother got from Grandma
[00:10:59.730]and I got from Grandpa or vice versa.
[00:11:03.010]And so the more distantly related the two people,
[00:11:07.610]the less overlap there would be,
[00:11:09.320]so this overlap is gonna be maximal
[00:11:11.860]in the case of close relatives like siblings.
[00:11:16.840]All right, so these chunks of ancestry, then,
[00:11:20.140]which we refer to as identical by descent,
[00:11:23.240]these are what we're looking for,
[00:11:24.750]and this is how we're gonna use these chunks
[00:11:28.090]to estimate ancestry, and to also
[00:11:32.660]try and understand how these DNA sequences
[00:11:34.800]affect things we're interested in, phenotypes,
[00:11:37.300]traits, health, wealth, height, what have you.
[00:11:42.140]So the way that we do this is by looking at variation
[00:11:44.630]in the DNA sequences, all right?
[00:11:46.410]So this is an example of a kind of information
[00:11:50.840]that we're looking at that we're storing in these computers.
[00:11:53.450]So what we have here is imagine a DNA sequence
[00:11:55.750]from a single gene or a single spot on a chromosome.
[00:11:58.530]These ellipses indicate large gaps.
[00:12:01.330]So this stretch might in fact
[00:12:02.700]be millions of base pairs long,
[00:12:04.010]and we have here sequences from 12 different individuals.
[00:12:08.660]And so what may not be obvious but I will highlight
[00:12:11.250]is that at some of these places,
[00:12:13.500]individuals have different bases.
[00:12:16.553]So at this position, some individuals
[00:12:18.520]have a T and others have a C.
[00:12:20.950]And this is one kind of variation that exists in the genome,
[00:12:24.590]this is the most common kind of variation.
[00:12:27.720]We call these variants
[00:12:29.480]single nucleotide polymorphisms or SNPs.
[00:12:33.950]And we can use, then, these SNPs
[00:12:36.620]to identify regions of identity by descent.
[00:12:39.630]So for example, we could use polymorphisms here
[00:12:43.440]to instead of simulating this, we could actually sequence
[00:12:46.660]my genome and my brother's genome and my dad's genome
[00:12:49.377]and infer this kind of overlap.
[00:12:54.730]All right, so
[00:12:57.950]we as a species have genetic variation,
[00:13:02.480]and on average, your maternal
[00:13:05.470]and your paternal chromosomes differ
[00:13:08.550]at one in 1000 bases,
[00:13:11.080]or about 3 million SNPs across the genome.
[00:13:14.160]So one way to say that is that you are likely to be
[00:13:16.450]heterozygous, that is carry two different sequences,
[00:13:20.410]two different bases at something like
[00:13:22.580]3 million positions in your genome.
[00:13:26.750]All right, so what does this variation do?
[00:13:30.380]Most of it, at least SNPs, does nothing.
[00:13:33.040]So most of these bases do not affect
[00:13:35.720]any trait in any detectable way,
[00:13:38.540]but some of them obviously do.
[00:13:40.180]And so that genetic variation is
[00:13:44.230]of singular importance to the medical community,
[00:13:46.680]and we have made tremendous strides over the last
[00:13:49.530]30, 40 years in understanding the genetic reasons
[00:13:53.240]why people get sick,
[00:13:54.870]understanding the bases for genetic diseases,
[00:13:57.540]and I would say only more lately,
[00:14:00.540]but in the last decade or so,
[00:14:01.900]we've started to make significant progress
[00:14:03.580]in turning that information into actual treatments.
[00:14:07.610]So, an example of, maybe the most famous example,
[00:14:11.870]is a disease called cystic fibrosis.
[00:14:14.780]So cystic fibrosis is a very simple genetic disease,
[00:14:17.520]it is caused by mutations in one gene,
[00:14:20.530]the CFTR gene, which encodes a transmembrane receptor,
[00:14:25.190]it's an ion channel, ion chlorite, ion transporter,
[00:14:29.190]and defects in this gene prevent cells in the lung
[00:14:32.330]from maintaining proper osmotic balance,
[00:14:34.630]and that leads to mucus buildup and infections
[00:14:36.680]and all of the symptoms of cystic fibrosis.
[00:14:41.387]It was one of the first genes and disease causing mutations
[00:14:44.320]discovered, it was sequenced in 1989,
[00:14:47.310]and then two weeks ago,
[00:14:49.740]the FDA just approved a triple drug therapy
[00:14:54.600]that seems like it's got a lot of promise,
[00:14:58.270]it seems like it's gonna be very effective.
[00:15:00.510]Obviously you have to wait and see
[00:15:02.440]10 years out, but it's very promising.
[00:15:05.370]And this drug therapy
[00:15:07.780]specifically targets the defects associated
[00:15:11.230]with the mutations that cause cystic fibrosis.
[00:15:14.370]So it's very unlikely that this therapy
[00:15:16.770]would have been developed in the absence
[00:15:18.470]of the information about this gene.
[00:15:21.120]But note that it was 30 years from discovery of the gene
[00:15:25.470]to a treatment, so this stuff doesn't come about overnight.
[00:15:30.220]Another application of DNA sequence information
[00:15:34.970]to medical treatment is
[00:15:38.362]an emerging area called pharmacogenomics,
[00:15:40.590]and that describes the interactions between
[00:15:43.670]genetic variation in humans and response to drug treatment.
[00:15:49.150]And so one example of this involves a gene called CYP2D6,
[00:15:54.890]it doesn't matter what that is.
[00:15:56.790]It's a gene that mostly has its action in the liver
[00:16:00.000]and detoxifies things that we put into our body.
[00:16:03.550]And so that includes small molecule drugs
[00:16:06.170]that are taken for therapy.
[00:16:08.740]And this gene, we have a number of different enzymes,
[00:16:13.050]genes in our body that do this kind of detoxification.
[00:16:16.030]This particular one is thought to metabolize
[00:16:19.040]about 25% of all of the drugs that are prescribed,
[00:16:22.650]and over the counter as well.
[00:16:25.060]And it turns out that this gene
[00:16:26.470]is incredibly variable among people.
[00:16:29.810]And so there are more than 70 different variants,
[00:16:32.300]and that those variations are not just
[00:16:34.480]A's versus T's or C's versus G's,
[00:16:36.970]it turns out that at this gene,
[00:16:38.240]one of the ways that humans are variable
[00:16:39.810]is how many copies of this gene you have.
[00:16:43.020]So some people have on a given chromosome, one copy,
[00:16:46.650]some people have copies that are more or less defective,
[00:16:49.570]some people have zero copies, it's gone,
[00:16:51.980]those people are more or less fine.
[00:16:54.760]Some people have two, and then some people
[00:16:56.410]have up to 14 copies.
[00:16:58.660]And copy number variation at this gene
[00:17:01.610]has a strong effect on how rapidly you metabolize
[00:17:04.960]the drugs that are the target of this gene.
[00:17:07.800]So having too few copies
[00:17:10.020]leads to a buildup of the drug and can cause toxicity.
[00:17:14.460]Having many copies can cause the drug
[00:17:17.620]to be metabolized too quickly to be effective,
[00:17:19.600]or can also instead cause side effects.
[00:17:23.840]And so there are now, if you, for example,
[00:17:26.480]go to the Mayo Clinic's website,
[00:17:28.730]they have a whole bunch of tests that you can do
[00:17:32.800]to look for interactions between CYP2D6
[00:17:36.780]or related gene CYP2C19
[00:17:40.250]and a whole bunch of different drugs.
[00:17:43.020]And if you have a particular genotype that's CYP2D6,
[00:17:47.830]they recommend that codeine not be used
[00:17:50.870]because you're gonna metabolize it too quickly
[00:17:52.780]or too slowly, and they suggest
[00:17:54.680]a different drug for that application.
[00:17:58.570]So this has a lot of promise
[00:18:01.690]for reducing side effects, adverse drug reactions,
[00:18:06.830]and just sort of basically understanding
[00:18:09.030]how our bodies deal with these chemicals
[00:18:11.620]that we put in them, often for very important,
[00:18:14.700]necessary uses.
[00:18:20.370]A third medical application of sequencing is for cancer.
[00:18:24.780]So cancer is a disease
[00:18:29.160]that results from ourselves sort of deciding
[00:18:33.530]that they're gonna strike out on their own.
[00:18:35.760]They don't need to listen to our bodies anymore.
[00:18:38.470]And as a result, they cause problems
[00:18:40.910]when they invade other tissues.
[00:18:42.970]And we've discovered that cancer
[00:18:46.320]is maybe hundreds of different diseases,
[00:18:48.710]at least at the genetic level.
[00:18:50.750]And at one level, that's very frustrating
[00:18:53.450]because that means that there's not probably going to be
[00:18:56.000]a cure for all cancers.
[00:18:59.310]But on the other hand,
[00:19:00.860]once we've started to disentangle the various
[00:19:03.400]genetic mechanisms that lead to cancer,
[00:19:05.720]that has led to some very promising therapies
[00:19:09.220]that are specific, but in the context
[00:19:12.560]of a particular tumor type can be very effective.
[00:19:15.550]So one example of this are non-small cell lung cancers.
[00:19:21.080]There are three mutations
[00:19:23.280]that are very common among this type of cancer.
[00:19:25.870]They can be in an EGFR gene, this ALK gene,
[00:19:29.810]or this ROS1 proto-oncogene,
[00:19:32.300]and depending on which combination of these mutations
[00:19:36.030]a patient has, different drugs can be prescribed.
[00:19:40.120]So there are drugs that are specific for this cancer
[00:19:42.510]that are EGFR positive, there are drugs
[00:19:45.340]that are specific for ALK positive cells, and so on.
[00:19:48.770]And so by doing a biopsy and by sequencing the tumors,
[00:19:52.800]we can identify the specific mutations
[00:19:55.180]that led to that cancer in that patient.
[00:19:57.600]And again, it seems like targeting specific cancers
[00:20:01.960]with specific drugs seems to be much more effective than
[00:20:06.010]sort of broad spectrum chemotherapeutic approaches
[00:20:09.360]which try to kill all cancer cells, and as a consequence,
[00:20:13.500]take a big toll on non-cancer cells.
[00:20:18.730]All right.
[00:20:19.563]So, that's just the tip of the iceberg.
[00:20:22.800]I mean, there are lots and lots of
[00:20:26.100]applications of DNA sequencing to medicine,
[00:20:28.350]I haven't even talked about,
[00:20:29.490]and I don't have time to talk about gene therapy,
[00:20:32.450]but there are a couple of diseases that we now
[00:20:34.970]can go into cells and fix them
[00:20:36.940]and return those cells to the patient,
[00:20:38.820]and that seems to be a cure.
[00:20:41.980]So some early inherited leukemias, for example.
[00:20:47.270]But I wanna point out, highlight,
[00:20:49.330]that something like cystic fibrosis is genetically unusual,
[00:20:53.380]that is diseases that are as simple genetically
[00:20:57.140]as cystic fibrosis are definitely the exception, right?
[00:20:59.870]So cystic fibrosis, there is the CFTR gene
[00:21:02.020]that's on chromosome seven.
[00:21:03.370]If you have two copies of a mutant allele,
[00:21:06.210]you will get cystic fibrosis.
[00:21:07.990]If you have one copy of a non-mutant allele, you will not.
[00:21:11.090]And to a first approximation, that's the whole story.
[00:21:14.060]But that is very unusual.
[00:21:15.390]So for most health related traits,
[00:21:19.020]and indeed for most traits that we're interested,
[00:21:22.340]the situation is a lot more complicated, right?
[00:21:24.840]So for most things, diseases that you might get,
[00:21:28.630]phenotypes that you might show,
[00:21:30.410]that is regulated, controlled by lots of genes
[00:21:33.880]on lots of chromosomes all over the genome.
[00:21:37.150]And the effect of each of these variants
[00:21:39.710]on that phenotype is small.
[00:21:42.460]And sometimes it matters whether you get
[00:21:44.640]this one and this one or this one and this one,
[00:21:46.540]so they interact, and they interact in ways
[00:21:49.190]that are often very unpredictable.
[00:21:51.280]So it's not as if we can simply sequence everyone's genome
[00:21:55.590]and know exactly how their life is going to go
[00:21:59.630]with respect to health, or in fact, anything else.
[00:22:03.880]So we have methods now, very sophisticated,
[00:22:07.510]powerful methods for understanding the genetic basis
[00:22:11.830]of things that are more complicated than cystic fibrosis.
[00:22:15.630]And these can be lumped under the term
[00:22:18.770]genetic association studies.
[00:22:20.680]So the concept behind these is very simple.
[00:22:24.050]We have variation in sequences, we have these SNPs,
[00:22:28.650]and we simply look at the sequences
[00:22:32.540]over the entire genome for
[00:22:35.010]let's say individuals who are healthy,
[00:22:37.040]in individuals, if we're thinking about a disease,
[00:22:38.830]who are affected.
[00:22:40.720]And what we look for are polymorphisms
[00:22:43.850]where they are enriched among affected individuals
[00:22:47.430]relative to non-affected individuals.
[00:22:50.110]So in this cartoon example, you may notice here,
[00:22:54.180]the C allele at this site
[00:22:56.510]is perfectly associated with the phenotype,
[00:23:00.650]so all of the affected individuals have C,
[00:23:02.950]and all of the non-affected individuals have T.
[00:23:06.020]And so in this context, we would say
[00:23:08.310]that this SNP likely is not causal
[00:23:12.600]but is likely very close to a causal polymorphism
[00:23:16.890]that leads individuals to get this condition.
[00:23:21.650]Now, this is obviously a made up cartoon
[00:23:24.640]and it's never that clean.
[00:23:28.540]In fact, we now have studies that are quite large
[00:23:31.690]so we can detect very subtle effects.
[00:23:34.630]So it may be that the frequency of the C allele
[00:23:37.170]among affected individuals is 10%,
[00:23:40.510]but the frequency of the C allele among
[00:23:42.670]non-affected individuals is 80%,
[00:23:45.060]and nonetheless that may be a sufficient difference
[00:23:48.410]for us to be able to detect that, and it is real.
[00:23:51.130]But it means that that variant
[00:23:53.220]contributes very little to the outcome by itself.
[00:23:59.010]So to highlight this,
[00:24:02.460]this figure's a little dark, but I really love this figure.
[00:24:05.730]This is showing a trait
[00:24:09.300]that is more typical of a lot of traits in humans,
[00:24:12.310]and that is height.
[00:24:13.800]So height, like many other traits in humans,
[00:24:16.070]is not on or off, it's a continuous trait.
[00:24:19.250]And so this is, I believe, a statistics class
[00:24:22.920]that has lined up according to their height.
[00:24:26.170]So the shortest individuals are here,
[00:24:28.110]the tallest individuals are here, I have to guess
[00:24:30.640]that there was someone in the class who was 4'10
[00:24:33.150]but decided not to show up on the day of the picture,
[00:24:36.110]perhaps understandably.
[00:24:38.580]And so this highlights the variation that we see
[00:24:42.400]in populations, continuous variation.
[00:24:44.750]And height is actually a trait
[00:24:46.760]that has been studied intensively, and then probably
[00:24:49.840]the biggest association study to date
[00:24:52.110]has been done on human height.
[00:24:53.950]And so this has been done a lot of times,
[00:24:55.970]there have been some great meta analyses,
[00:24:57.790]and we now think that human height is detectably influenced
[00:25:02.010]by over 3000 genes across the genome.
[00:25:05.550]So it is truly polygenic.
[00:25:07.290]There are tons of sites everywhere that affect height.
[00:25:10.730]Each one only changes an individual's height
[00:25:13.870]by a millimeter or a fraction of a millimeter,
[00:25:16.930]and then the total height, then,
[00:25:19.700]is the sum of all of those effects.
[00:25:21.650]But even when we sum all of those up,
[00:25:23.440]that together accounts for something like a quarter
[00:25:26.040]of the variation among individuals.
[00:25:28.340]So in this room, there's variation in height,
[00:25:30.870]and maybe a quarter of that variation
[00:25:33.150]can be explained by the genes that those individuals have.
[00:25:35.960]The rest of that variation we attribute to the environment,
[00:25:39.820]which means something other than the genes.
[00:25:42.150]So that would be what you ate growing up,
[00:25:44.890]that would be any number of other
[00:25:47.440]non-genetic influences that determine your final height.
[00:25:52.640]The reason I really love this figure is it illustrates
[00:25:54.732]the multifarious nature of genetic control of traits,
[00:25:58.990]because this is a continuous trait,
[00:26:01.130]there's 3000 different loci
[00:26:03.870]are in this population that are affecting height,
[00:26:06.450]but there is also one gene
[00:26:07.900]that has an enormous effect on height.
[00:26:11.096]So all of the men here are dressed in blue,
[00:26:13.570]all of the women are wearing white,
[00:26:15.200]and you can see quite clearly the huge effect
[00:26:17.350]of having a Y chromosome.
[00:26:19.350]And so this is true not just of height,
[00:26:20.980]but this is true of many other traits.
[00:26:23.950]Sex has a big effect,
[00:26:25.560]biological sex has a big effect on health.
[00:26:27.820]And so this is just to keep in mind,
[00:26:30.570]there are big things and there are small things,
[00:26:32.670]and it's not one or the other.
[00:26:40.282]So this idea has been encapsulated in this review figure.
[00:26:47.060]There are rare disease alleles that cause Mendelian disease
[00:26:50.430]that have certain large effects, but those are very rare.
[00:26:53.660]So very few diseases are like cystic fibrosis.
[00:26:57.560]There are many common diseases,
[00:27:01.030]but those tend to have a genetic architecture
[00:27:05.430]where they are controlled by many, many genes
[00:27:07.890]and each gene contributes a very, very small effect.
[00:27:10.640]And so obviously, these kinds of traits,
[00:27:14.120]these kinds of diseases are much trickier to work with
[00:27:17.400]in terms of genetic prediction or in terms of thinking about
[00:27:20.320]should we use genetic information to target therapies.
[00:27:27.650]All right.
[00:27:29.320]It gets worse than that
[00:27:32.300]with respect to predicting human phenotypes,
[00:27:35.490]human traits based on genetics.
[00:27:38.100]And that's because so much of our biology
[00:27:42.030]is affected by our behavior.
[00:27:44.460]And our behavior, while there is definitely
[00:27:47.220]a genetic component to lots of our behavior,
[00:27:49.510]most of our behavior is learned
[00:27:51.920]and culturally transmitted through families
[00:27:54.480]and through our society.
[00:27:56.000]And so this was revealed by a really cool study last year
[00:28:00.820]that was done by Ancestry.com, which is,
[00:28:03.870]I'll talk about them a little more later,
[00:28:05.590]but one of these sites where you can upload,
[00:28:08.700]where you can get a tests that will
[00:28:11.500]genotype you across the genome to look for
[00:28:14.470]regions that indicate where your ancestors may have lived,
[00:28:17.300]and then Calico which is Google's aging research company.
[00:28:22.680]So they combined the information,
[00:28:25.970]and so Ancestry both has DNA, but also individuals
[00:28:28.790]can upload information that they got simply from pedigrees.
[00:28:32.570]So, taking family histories.
[00:28:34.080]And so they were able to put together
[00:28:35.640]all of this information and came up with a single pedigree
[00:28:39.420]that consisted of 400 million individuals
[00:28:42.520]that went back, I don't know, 10 generations or so.
[00:28:46.080]And so they had a little bit of demographic information
[00:28:50.440]including how long these people lived.
[00:28:52.720]So they thought, "All right, let's see if we can do a study
[00:28:55.007]"estimating the heritability of longevity,"
[00:28:58.280]that is, and for this study they didn't use
[00:29:00.690]any DNA information, they just used families.
[00:29:03.460]So the idea here is
[00:29:06.210]if something runs in a family
[00:29:08.850]then that could be because it's genetic,
[00:29:10.730]that could be because it's culturally transmitted,
[00:29:12.970]but either way, we would call that the heritability,
[00:29:15.640]that is the correlation between relatives.
[00:29:17.910]So they had to, it's a complicated data set
[00:29:20.490]and they had to do a whole bunch of stuff
[00:29:22.580]to make sure they were doing it right,
[00:29:24.730]including sort of stratifying by decade,
[00:29:26.720]which the heritability of longevity
[00:29:28.910]shows a lot of variation across decades.
[00:29:31.430]No idea why that is, it's kind of interesting.
[00:29:34.440]But they found that the sort of 20 to 30% range
[00:29:39.870]seemed to be the effect of heritability on longevity,
[00:29:43.786]that is about 20 to 30% of the variation in lifespan
[00:29:47.100]could be predicted by which family you are in.
[00:29:51.300]And so they could see that it was pretty consistent
[00:29:54.030]regardless of the degree of the relatives, right?
[00:29:56.170]So for siblings, you predict a certain correlation
[00:29:59.670]based on their genetic similarity.
[00:30:01.370]For cousins, you predict a lower correlation
[00:30:05.100]because they share fewer genes.
[00:30:07.430]But if you correct for that, the heritability should be,
[00:30:10.400]again, they have estimated roughly 20 to 30%.
[00:30:13.510]Okay.
[00:30:15.020]But they noticed a couple of other patterns
[00:30:18.130]that are a little hard to square with this.
[00:30:21.030]One is that spouses tended to have
[00:30:24.910]more similar lifespans than siblings.
[00:30:28.790]So you share on average half of your genes
[00:30:31.100]with your siblings.
[00:30:33.020]On average, how many genes do you share with your spouse?
[00:30:37.360]Hopefully zero, right.
[00:30:41.800]Yet the correlation between
[00:30:45.070]spouse lifespan is much higher than siblings.
[00:30:49.660]Now, notice there are four kinds of siblings here,
[00:30:52.660]or three, rather.
[00:30:54.040]Brother brother, sister sister, and then brother sister.
[00:30:59.200]And you notice the effect of sex here,
[00:31:01.270]where brother sister lifespans are less correlated
[00:31:04.610]than same sex siblings, so they're comparing this
[00:31:08.327]to the opposite sex siblings, which makes sense.
[00:31:11.950]But still, this suggests that there is
[00:31:15.320]a strong effect of shared environment,
[00:31:17.690]and this isn't actually that surprising.
[00:31:21.960]You obviously get behaviors from the people you live with.
[00:31:26.010]And so lifestyle choices probably account
[00:31:29.730]for this correlation between spouses.
[00:31:32.970]However, they also noticed
[00:31:35.960]that when they were looking at correlations
[00:31:38.300]between cousins and cousins-in-law,
[00:31:42.360]so cousins, you share some genes with them,
[00:31:44.420]cousins-in-law you do not.
[00:31:46.670]They noticed that the correlations were pretty similar.
[00:31:50.750]So in-law relatives have lifespan correlations
[00:31:53.570]that are almost as strong as blood relative correlations.
[00:31:57.410]So you might think that that, again,
[00:31:59.570]is evidence for shared culture, for cultural transmission
[00:32:03.440]of traits, behaviors that lead to differences in lifespan.
[00:32:08.320]But they point out that for some of these cousins,
[00:32:12.300]these are relatives who
[00:32:15.040]traditionally don't share a household.
[00:32:17.590]So these are distant enough that,
[00:32:19.540]at least in the last few generations,
[00:32:21.900]they probably weren't living together.
[00:32:23.600]And so they raised an alternative explanation
[00:32:26.530]for this correlation, which was assortative mating.
[00:32:30.470]That is the idea that individuals who have behaviors,
[00:32:33.670]traditions, practices that lead to a long life span
[00:32:37.520]tend to marry unrelated individuals
[00:32:39.810]who have similar cultural behavioral practices.
[00:32:45.040]And vice versa for individuals
[00:32:46.960]who tend to have shorter lifespans.
[00:32:49.230]And that, if you think about
[00:32:51.160]sort of how our societies are stratified
[00:32:53.200]and how we pair up, that also makes good sense.
[00:32:57.510]But that kind of assortative mating
[00:32:59.840]is definitely a problem for genetic association studies.
[00:33:03.840]Sort of the first, one of the first assumptions
[00:33:05.980]you need to make for genetic associations
[00:33:07.950]and for looking at heritability across pedigrees
[00:33:10.330]is that individuals who come together to produce
[00:33:13.510]the next generation don't share an environment.
[00:33:15.930]And this suggests, actually, a quite strong
[00:33:18.340]sharing of environment.
[00:33:19.780]Now, this may be that lifespan is the worst trait for this
[00:33:23.080]because lifespan is gonna be the sum of a lot of behaviors
[00:33:26.530]over an individual's entire life,
[00:33:29.090]so maybe for other traits it wouldn't be as bad,
[00:33:31.340]but it still highlights the danger of
[00:33:35.610]putting too much of this on the DNA itself.
[00:33:41.040]All right, so, one moment.
[00:33:57.580]A lot of the DNA sequencing that has been done
[00:33:59.600]in the last decade or so has been done by scientists,
[00:34:03.150]by medical practicers, by clinicians.
[00:34:07.330]But there has recently been a real explosion
[00:34:10.090]in the direct to consumer genetic testing industry.
[00:34:15.020]And so 23andMe and Ancestry are two of the biggest hitters.
[00:34:19.740]These are companies where, as I'm sure all of you know,
[00:34:23.080]you send off a DNA sample, you spit in a tube
[00:34:25.570]and they genotype you at over 700,000 SNPs,
[00:34:31.760]so, polymorphisms in your genome.
[00:34:34.470]And then will tell you, first of all,
[00:34:39.720]in their database of other individuals who have done this,
[00:34:43.220]do you have any relatives, can you find close matches,
[00:34:46.270]can you find more distant matches?
[00:34:48.550]They can tell you a little bit about
[00:34:51.070]where your ancestors may have lived.
[00:34:53.410]And then they are starting to get into,
[00:34:57.130]in fits and starts,
[00:35:00.430]giving you information about SNPs that, in some contexts,
[00:35:03.730]have been linked to phenotypes,
[00:35:06.010]including health-related phenotypes.
[00:35:09.220]This has been sort of a tussle with the FDA
[00:35:11.560]because 23andMe does not wanna be
[00:35:15.030]a genetic counseling service.
[00:35:18.250]So they weren't doing it and now they're doing it again
[00:35:20.740]and we'll see how it goes in the future.
[00:35:23.250]In addition, there are a number of companies,
[00:35:27.590]websites, databases that don't actually do the sequencing,
[00:35:32.240]but when you use one of these kits,
[00:35:35.960]you can download your data.
[00:35:37.820]And then you can upload it again to a place
[00:35:41.070]like GEDmatch or FamilyTreeDNA.
[00:35:46.070]And these, again, the point of these
[00:35:49.560]are to use your DNA to look for relatives,
[00:35:52.960]and to look to see if you have
[00:35:56.050]alleles that indicate something about
[00:35:58.310]where your recent ancestors might have lived.
[00:36:05.240]So this works because
[00:36:09.880]over recent human histories, over deeper human history
[00:36:13.340]we kind of spread out over the globe,
[00:36:15.240]and then until recently, didn't move too, too much.
[00:36:19.210]And so that meant that there are alleles
[00:36:21.820]that got to different places
[00:36:23.530]or reached different frequencies in different places,
[00:36:26.790]and those alleles, although they are a minority
[00:36:29.880]of the polymorphisms in the genome,
[00:36:31.590]there are plenty of them, and so those can be used
[00:36:34.610]to find out where your ancestors
[00:36:37.180]that contributed that DNA may have resided.
[00:36:39.470]So this is a fairly famous figure from a famous paper now
[00:36:42.910]in 2008 looking at genetic variation in Europeans.
[00:36:47.180]And so what's plotted here is a
[00:36:50.700]statistical summary of the genetic data,
[00:36:54.240]principal components, so these data have been summarized
[00:36:57.070]down into two major axes that capture
[00:36:59.680]most of the variation between individuals
[00:37:02.710]in their DNA sequences.
[00:37:04.890]And they are orthogonal to each other,
[00:37:07.810]and then lo and behold,
[00:37:11.410]each individual here, so each pair of letters
[00:37:15.330]is an individual, and their location on this axis,
[00:37:20.550]on this plot is based on their genetic variation,
[00:37:23.750]and then their color of the point or the letters
[00:37:27.010]is where they reside.
[00:37:30.930]And you can see that there is a striking relationship
[00:37:34.160]between their genetic variation on this plot
[00:37:37.930]and their geography of Europe.
[00:37:40.020]So the colors here match,
[00:37:42.110]these are individuals from Spain and Portugal.
[00:37:44.920]These are individuals from Russia,
[00:37:46.780]Finland, Ireland, Great Britain.
[00:37:49.600]And so this striking correspondence
[00:37:52.250]between genetic variation and geography
[00:37:55.320]exists because for a number of generations,
[00:37:58.490]individuals didn't move around too much.
[00:38:01.090]And so those alleles that were at higher frequency
[00:38:03.920]in this spot stayed there,
[00:38:05.940]and consequently those individuals,
[00:38:08.450]then when they emigrated to the United States
[00:38:11.050]and had children and grandchildren, great grandchildren,
[00:38:13.750]you can see in the DNA sequences of those individuals
[00:38:17.930]the signature of where they came from.
[00:38:21.830]So these databases have really been amazing
[00:38:25.430]for finding relatives.
[00:38:27.660]They've been used a lot by people who were adopted
[00:38:30.300]to find their biological parents.
[00:38:33.050]But they have also proven to be a boon for law enforcement.
[00:38:37.950]And so this was really dramatically shown
[00:38:41.280]recently with the identification
[00:38:45.000]of the Golden State Killer, a serial killer
[00:38:47.130]who terrorized California for decades.
[00:38:50.090]And so the way that they found him
[00:38:53.390]was there was DNA that was left at the crime scenes
[00:38:56.540]way back then, and law enforcement uploaded
[00:39:00.750]that DNA to this database,
[00:39:03.850]which had, at the time, about a million profiles in it.
[00:39:08.350]And they were able to find a partial match.
[00:39:10.870]So this was a relative who was a,
[00:39:13.970]they shared a great great great great grandfather,
[00:39:17.090]so a third or fourth level cousin.
[00:39:21.050]And that, they were then able to build from that
[00:39:23.750]a family tree, so all of the individuals
[00:39:26.430]who are connected between the DNA found at the crime scene
[00:39:29.470]and this individual in the database,
[00:39:31.530]and that was like 1000 people.
[00:39:33.440]And from those 1000 people, they were then able to use
[00:39:37.690]age and sex and where they lived
[00:39:40.670]to narrow it down to very few candidates.
[00:39:44.070]And ultimately, the alleged
[00:39:48.910]individual was identified by
[00:39:53.990]going to his environs
[00:39:56.790]and taking DNA from his trash or his car door handle.
[00:40:02.990]And that then matched the DNA
[00:40:04.700]that was taken from the crime scene.
[00:40:07.040]So, this has revealed a lot of
[00:40:11.630]really sticky issues regarding genetic privacy.
[00:40:16.170]And so this is basically,
[00:40:20.290]it was said and I agree with this, the wild wild west.
[00:40:24.150]We don't know what the rules are,
[00:40:26.420]we don't know what the rules should be,
[00:40:28.110]and the rules are changing under our feet.
[00:40:30.980]So shortly after this,
[00:40:34.950]GEDmatch changed their terms
[00:40:37.540]so that you now have to voluntarily opt in
[00:40:41.380]if you want your DNA to be publicly available.
[00:40:44.200]They also said we don't want law enforcement
[00:40:46.840]pretending to be someone uploading fake profiles
[00:40:50.480]in order to find people.
[00:40:52.080]Both Ancestry and 23andMe take,
[00:40:55.330]or at least have said they've taken,
[00:40:56.950]strong precautions to prevent this sort of thing,
[00:40:58.980]so individual's DNA profiles for those companies
[00:41:02.900]should be unavailable to law enforcement or to the public.
[00:41:08.800]But I think it's really worth questioning
[00:41:14.520]what level of genetic privacy do we expect is ideal,
[00:41:19.680]is the right balance between protecting individuals
[00:41:22.690]and allowing law enforcement
[00:41:25.080]access to what is undoubtedly a very powerful tool?
[00:41:29.090]So, one concern that's come up
[00:41:31.600]are ways in which these databases can be attacked.
[00:41:35.950]And so just last week, couple weeks ago,
[00:41:39.840]two papers were published showing ways that you can
[00:41:43.310]identify individuals, anonymous individuals in a database
[00:41:47.390]if you upload lots of pretend genomes,
[00:41:51.800]pretend DNA that has particular
[00:41:54.880]SNPs that you know that you're looking for.
[00:41:57.320]And as of the time that these were published,
[00:42:00.500]they contacted Ancestry and GEDmatch
[00:42:02.957]and some of these other databases,
[00:42:04.580]and the ways that they described to do this,
[00:42:08.000]they could do it.
[00:42:08.880]So there are safeguards that these companies,
[00:42:11.520]that these databases could take to prevent this,
[00:42:13.940]but they have not done so yet.
[00:42:18.780]It's also, I think, worth considering that last step
[00:42:22.690]in identifying the Golden State Killer.
[00:42:25.230]We leave DNA everywhere.
[00:42:28.280]Like, I leave this room, you could come up to this podium
[00:42:31.380]and prove that I gave this talk, right?
[00:42:33.770]We shed cells, those cells contain our DNA,
[00:42:37.260]and it goes away eventually,
[00:42:39.410]but it would be trivial for law enforcement or anyone else
[00:42:42.900]to go to your garbage, to go to your car,
[00:42:45.720]to go to something that you've touched and collect your DNA.
[00:42:49.910]So is that part of the privacy,
[00:42:52.810]am I entitled to people staying away from the cells
[00:42:55.513]that I'm shedding into the environment?
[00:42:58.160]And then as of two hours ago,
[00:43:00.970]the New York Times reported that a judge issued a warrant
[00:43:06.050]to a police officer for GEDmatch.
[00:43:10.610]So now under this warrant, it doesn't matter
[00:43:14.680]whether individuals have opted out or not,
[00:43:17.510]it doesn't matter whether the owners of GEDmatch
[00:43:20.020]said you can't do that, they are compelled
[00:43:22.680]by this warrant to comply with law enforcement.
[00:43:26.190]And so it may be that in the coming months,
[00:43:29.890]it seems like if the public and government
[00:43:32.840]sides on the side of law enforcement,
[00:43:35.040]it may be that none of these databases are secure.
[00:43:39.720]And it's worth considering also,
[00:43:42.130]you may not have done one of these tests,
[00:43:45.110]you may not have uploaded your profile to GEDmatch,
[00:43:49.710]but has anyone even remotely closely related to you?
[00:43:53.950]If a third or a fourth cousin, like I have no idea
[00:43:55.890]who my third or fourth cousins are.
[00:43:57.650]But if they've put their data up there,
[00:43:59.560]then you can be found.
[00:44:01.890]So this is really, we're at a critical moment,
[00:44:05.230]I think, for these privacy concerns.
[00:44:09.390]Okay, I wanna end here by talking a little bit about
[00:44:12.870]the nature of human genetic variation.
[00:44:17.110]There are differences between populations,
[00:44:19.710]so we originated in Africa, there we are,
[00:44:23.040]and spread all over the world.
[00:44:25.000]We share a common ancestor in Africa
[00:44:27.040]a few hundred thousand years ago.
[00:44:29.640]Since then we have diverged
[00:44:31.600]from each other in various phenotypes,
[00:44:34.580]and those tend to be the ones that we often focus on.
[00:44:38.970]But while those differences are real and genetically based,
[00:44:44.230]they are actually the result of a very small fraction
[00:44:47.380]of the variation that's in the human genome.
[00:44:49.930]So what's shown here are pie charts
[00:44:52.800]indicating within each of these populations
[00:44:55.790]how the genetic variation
[00:44:57.120]is apportioned across populations.
[00:44:59.740]So in these pie charts, the gray colors are variants,
[00:45:04.120]polymorphisms that are shared either across all continents
[00:45:07.200]or across multiple continents.
[00:45:08.500]So these are people in Yoruba have A's and T's,
[00:45:13.900]people in Finland have A's and T's at that site.
[00:45:17.650]So those are polymorphisms
[00:45:19.270]that are shared across all humans.
[00:45:22.160]And then the colored pies are either private to a continent,
[00:45:26.830]so Africa or Asia or Europe or the Americas,
[00:45:30.000]or private to a population within,
[00:45:32.230]and you can see that in most of these circles,
[00:45:35.530]it is dominated by the gray colors, right?
[00:45:37.710]So, particularly in Europe,
[00:45:39.510]very small fraction of the variants
[00:45:41.370]are private to those populations.
[00:45:44.390]Now those, again, are responsible
[00:45:47.010]for some phenotypic differences that we have
[00:45:49.450]traditionally focused on and thought of as significant.
[00:45:52.490]So it's not to say that there aren't phenotypic differences,
[00:45:55.180]just that most of that variation is not specific
[00:45:58.640]to a population, it's shared by all people.
[00:46:03.490]Another couple of points I wanna make about genealogies
[00:46:08.380]that I think are, at least were
[00:46:09.530]a little counterintuitive to me,
[00:46:11.910]thinking about all of our ancestors, on average,
[00:46:14.170]I said, one over two to the K of our genome
[00:46:16.860]is inherited from each ancestor
[00:46:18.590]that lived K generations ago.
[00:46:21.160]But that average, again, is misleading
[00:46:23.100]because of the stochastic nature
[00:46:24.510]of transmission of chromosomes.
[00:46:27.180]So, right, remember here,
[00:46:29.860]for chromosomes 20 through 14, or 13 in this example,
[00:46:33.960]I inherited either from Dad, from his mom
[00:46:36.540]or from his dad, but not pieces of both.
[00:46:40.670]And as a result, this is what your genome
[00:46:43.760]actually looks like in these ancestors.
[00:46:47.470]And so you can see that right at about generation 10,
[00:46:51.740]in fact, most, the majority of your genealogical ancestors,
[00:46:57.530]you carry none of their DNA.
[00:47:00.100]And the further back you go, the worse that gets,
[00:47:02.880]or the more extreme that gets.
[00:47:05.305]So the further back you go, the more ancestors you have,
[00:47:08.510]but most of them will have contributed
[00:47:10.490]none of the DNA in your cells.
[00:47:12.940]Now, that doesn't make them any less your ancestors,
[00:47:15.720]that doesn't mean that they weren't instrumental
[00:47:18.860]in your family being who it is,
[00:47:21.710]like all of them had to survive and have offspring
[00:47:23.790]for you to be here, but not all of them
[00:47:26.840]transmitted their DNA to you.
[00:47:29.340]And in fact, this is, again, a simulation,
[00:47:31.680]so this is one example, but you can see here
[00:47:33.770]down on the bottom is the paternal line.
[00:47:36.850]So this is my father, my father's father,
[00:47:38.550]father's father, et cetera.
[00:47:40.290]And you can see in this example,
[00:47:42.500]after one, two, three, four, five, six,
[00:47:44.640]in the seventh generation,
[00:47:47.390]there's no material inherited from that individual.
[00:47:50.050]So because our society is patriarchal,
[00:47:52.490]that's probably where my name would come from,
[00:47:55.000]my Y chromosome maybe,
[00:47:57.670]these are the autosomes, not the Y chromosome,
[00:47:59.580]but other than that, I have no genetic
[00:48:02.340]inheritance from that individual.
[00:48:04.720]And you can see, actually, the same thing
[00:48:06.300]along the matra line happens at this generation.
[00:48:09.230]So we have a lot of genealogical ancestors,
[00:48:11.540]we have many fewer genetic ancestors.
[00:48:15.600]The flip side of that is that,
[00:48:19.641]remember we have two to the K ancestors
[00:48:23.010]K generations ago, that means 30 generations back,
[00:48:26.300]which is like 700, 800 years ago,
[00:48:30.430]each of us has 1 billion ancestors.
[00:48:33.290]But, of course, there weren't
[00:48:34.590]a billion people on the planet then.
[00:48:37.050]So what that means is that at some point
[00:48:38.830]you start to have repeat ancestors, right?
[00:48:41.160]So if this is a pedigree, here I am, my parents,
[00:48:43.500]my four grandparents, going back and back,
[00:48:45.990]at some point, there are individuals
[00:48:47.750]who are your ancestor multiple times over.
[00:48:50.950]And again, as we go further back in time
[00:48:53.410]and the number of genealogical ancestors
[00:48:56.270]continues to increase exponentially,
[00:48:58.860]it will be the case that
[00:49:01.880]most of your ancestors are repeat ancestors.
[00:49:04.830]And in fact, it's been mathematically determined
[00:49:07.540]that in a population of about 100,000,
[00:49:10.830]after 29 generations back,
[00:49:14.090]everyone who lived then and has any descendants
[00:49:18.060]is an ancestor for everyone in the population.
[00:49:21.530]Now, it's not clear what the human population size
[00:49:24.760]should be if it was historically 100,000
[00:49:27.140]or something different, but it's not gonna be far off.
[00:49:30.270]So when we think about Ancestry and you do your DNA test
[00:49:33.680]and you say, "My ancestors came from
[00:49:35.837]"Belarus or from Mexico,"
[00:49:38.900]that's your ancestors a few generations back.
[00:49:41.900]You only have to go back a few more
[00:49:44.200]before that doesn't really mean anything.
[00:49:46.770]You only have to go back 30 or 40 generations
[00:49:48.930]before it's clear that we all have
[00:49:50.670]the same set of ancestors.
[00:49:54.220]So, if we go back even further,
[00:49:57.720]DNA sequencing has taught us that
[00:50:01.190]there are some branches on the human tree
[00:50:03.210]that we thought sort of went extinct,
[00:50:05.020]but it turns out that they didn't really.
[00:50:07.340]So we've sequenced ancient DNA sequences
[00:50:10.200]from Neanderthal and Denisovan fossils.
[00:50:13.210]We've now gotten really good at getting DNA
[00:50:14.890]out of old bones that have been sitting in cold caves
[00:50:17.270]for a long time.
[00:50:18.460]And these are two branches on the human tree
[00:50:22.130]that left Africa many hundreds of thousands of years
[00:50:26.320]before the last exodus from Africa.
[00:50:29.630]And they colonized Neanderthals, colonized Europe and Asia,
[00:50:33.230]the Denisovans went through Asia
[00:50:34.860]and got all the way to New Guinea.
[00:50:37.440]And from fossils, we've known Neanderthals for a while,
[00:50:40.490]and we thought culturally they went extinct,
[00:50:42.580]they had a particular culture that sort of vanishes
[00:50:44.920]once anatomically modern humans moved in.
[00:50:49.210]But it turns out that this sequencing has also revealed
[00:50:51.930]that there were multiple introgression events,
[00:50:54.410]that is there was an interbreeding between the lineage
[00:50:57.670]that gave rise to modern humans and the lineage
[00:51:00.320]that gave rise to Neanderthals and Denisovans.
[00:51:03.240]And we know that this happened after we left Africa
[00:51:06.300]in Europe, because individuals for
[00:51:09.070]that have European and Asian ancestry, we find
[00:51:11.290]on average one to 4% of their DNA
[00:51:14.170]can be traced to a Neanderthal ancestor,
[00:51:16.500]but that is very, very small or absent
[00:51:19.230]among individuals with African ancestry.
[00:51:21.540]So this interbreeding happened outside of Europe
[00:51:24.240]after we had exited Europe.
[00:51:26.800]Similarly, we see evidence for three to 5% of inheritance
[00:51:31.030]from individuals who live in Melanesian Australia
[00:51:34.280]that is attributable to the Denisovan lineage.
[00:51:38.130]Okay, so obviously we haven't just been
[00:51:41.520]sequencing humans this whole time.
[00:51:43.750]We've been sequencing DNA
[00:51:44.950]from a whole bunch of other things.
[00:51:46.470]And I'd like to finish by pointing out
[00:51:48.200]that one of the things that's revealed is,
[00:51:50.160]again, sort of how similar we all are.
[00:51:52.980]So this is the amino acid sequence,
[00:51:56.350]so our DNA encodes information to make proteins
[00:52:00.130]and proteins have 20 letters instead of four.
[00:52:02.940]This is a protein called cytochrome c.
[00:52:06.790]And it is the cytochrome c, two copies of it from mouse,
[00:52:11.010]one from horse, one from human,
[00:52:12.480]and then one from budding yeast,
[00:52:14.690]so the organism that makes us beer and bread.
[00:52:17.010]And you can see quite clearly in this alignment
[00:52:20.060]and many of these positions in the protein are the same.
[00:52:23.340]And that's because this is the same gene.
[00:52:26.260]So this gene existed in the common ancestor of us
[00:52:30.030]and horses and mice and yeast about a billion years ago,
[00:52:34.410]and it did the same thing that it does today.
[00:52:37.330]It functions in the mitochondria,
[00:52:38.650]which you recall is this glowing firestorm thing
[00:52:41.410]in ourselves that gives us energy.
[00:52:43.670]And so over a billion years,
[00:52:46.200]this gene has been doing the same thing.
[00:52:48.900]It's been functioning in oxidative respiration,
[00:52:51.430]powering these cells, and so
[00:52:54.910]maybe one thing to take away from all of this DNA sequence
[00:52:58.030]is how similar we are not just to each other,
[00:53:00.900]but to all other organisms.
[00:53:02.270]So I'll just leave with this picture of my favorite
[00:53:05.380]organism, well, my two favorite organisms,
[00:53:09.650]which are, you know, share about 50% of their genes,
[00:53:12.700]and every day we're finding out are more similar
[00:53:15.000]to each other than we previously thought.
[00:53:17.600]And so with that, I will end.
[00:53:19.994](audience applauding)
[00:53:29.500]So it seems like you're afraid
[00:53:32.460]of the police coming in and using this kind of data.
[00:53:37.710]But what is your, it seemed like you were shying away
[00:53:41.290]from maybe what you perceive could happen.
[00:53:44.700]What is your worst fear with--
[00:53:48.250]Yeah, I was just sort of thinking about...
[00:53:54.714]I will say, I don't know.
[00:53:57.220]I have not yet, I think, imagined
[00:54:01.420]the worst thing that could be done.
[00:54:04.540]Certainly,
[00:54:08.665]I guess one perspective would be, well, so what?
[00:54:12.320]Maybe it doesn't matter if everyone knows
[00:54:14.320]what my DNA sequence is.
[00:54:16.530]I think in a world where,
[00:54:21.310]that would be great if that didn't matter.
[00:54:24.860]But I can imagine lots of nefarious purposes.
[00:54:29.010]So for example, I can imagine
[00:54:32.810]our current government deciding that
[00:54:35.100]the DNA that they are taking forcibly from immigrants
[00:54:39.750]and putting into a database reveals alleles
[00:54:43.340]that are associated with unlawfulness,
[00:54:46.500]which is garbage,
[00:54:48.010]but you could imagine them making that claim.
[00:54:51.120]You could imagine insurance companies
[00:54:53.460]harvesting DNA information to try and put you
[00:54:56.200]into certain risk groups.
[00:54:58.400]Aha, you've got this APOL1 allele,
[00:55:01.110]so we know that you're more likely to develop
[00:55:03.440]hypertension, so your premiums are gonna go up.
[00:55:06.650]So, I guess you could imagine
[00:55:13.720]people discovering, yeah, I mean, I guess
[00:55:17.480]I haven't imagined all of the bad things that could happen,
[00:55:20.960]and I'm not sure what they are yet.
[00:55:23.940]But I think it's maybe naive to think that
[00:55:26.130]someone won't find out a way to misuse this stuff.
[00:55:29.820]Thank you.
[00:55:36.421](audience member speaking faintly)
[00:55:56.090]I mean, there are many positives.
[00:55:59.190]Certainly, genetic medicine is real.
[00:56:03.250]And there are definitely,
[00:56:08.970]people are taking an approach
[00:56:11.780]that's sort of been used in animal breeding for a while now,
[00:56:14.770]to generate what are called polygenic risk scores.
[00:56:16.980]So this is the idea that for a trait that's like height
[00:56:19.430]that's controlled by lots of genes all over the genome,
[00:56:22.520]you just sequence someone's genome,
[00:56:24.910]figure out all of their polymorphisms added up,
[00:56:27.840]and determine their risk score for a particular trait.
[00:56:31.430]And it seems like for some common
[00:56:35.030]medical conditions like hypertension,
[00:56:37.370]like cholesterol level,
[00:56:39.950]polygenic risk scores work about as well
[00:56:42.690]as taking biomarkers, right?
[00:56:44.700]So if you go to the doctor
[00:56:46.460]and they take your cholesterol levels,
[00:56:48.030]if it's above a certain level, they may prescribe you
[00:56:51.724]a statin or some other medication.
[00:56:54.580]And it seems like if instead of taking that blood
[00:56:58.630]and doing that test, if they instead
[00:57:00.480]just looked at your genome,
[00:57:01.630]in some cases, that would be about as good.
[00:57:04.590]So that may be a real boon for predicting risk
[00:57:09.130]and for managing different propensities for diseases.
[00:57:12.740]I mean, it's still the case that for all of these
[00:57:14.970]conditions, you can look at your risk score
[00:57:18.400]or you can eat less bacon, and like it still works,
[00:57:22.120]it doesn't matter what your genes are,
[00:57:23.980]the behavior will change your risk.
[00:57:27.020]Certainly, I don't wanna pretend that
[00:57:31.240]the use of forensics is never appropriate
[00:57:34.670]in law enforcement, I mean,
[00:57:37.740]since the Golden State Killer, there are 70 cold cases
[00:57:40.820]that have been solved with the use
[00:57:42.680]of genetic genealogy detective work.
[00:57:46.180]So there are real upsides, and I think that's the challenge.
[00:57:49.700]If there were no upsides, it would sort of be easy
[00:57:52.860]to say, all right,
[00:57:56.875]we want to protect privacy,
[00:57:58.590]we want to focus on stuff other than DNA.
[00:58:01.780]But the challenges is that there are,
[00:58:03.870]it is so powerful in the medical realm,
[00:58:06.960]it is so powerful in the forensic realm.
[00:58:09.940]And so I think that leads to
[00:58:14.010]a danger that we're rushing into stuff
[00:58:16.190]before we have all the safeguards in place.
[00:58:22.240](audience member speaking faintly)
[00:58:36.430]I would say for most simple genetic diseases,
[00:58:40.230]it's been done.
[00:58:41.570]And a lot of it was done before
[00:58:43.490]the advent of these association studies,
[00:58:45.290]it was done with old fashioned linkage mapping
[00:58:47.770]where you would use pedigrees
[00:58:50.100]and you had like one marker per chromosome,
[00:58:54.390]so most of those, we've already identified those genes.
[00:58:58.950]I think in my limited understanding,
[00:59:02.300]treatment is gonna be specific
[00:59:04.450]for each gene, each disease.
[00:59:09.380]So for cystic fibrosis, this triple drug cocktail,
[00:59:15.320]one of the drugs improves the
[00:59:19.250]movement of the protein to the membrane.
[00:59:22.870]So, cystic fibrosis, I should say,
[00:59:27.000]there's one mutation that's super common.
[00:59:28.920]So, the majority of cases of cystic fibrosis
[00:59:31.900]are because individuals have at least one copy
[00:59:33.620]of this one mutation, but there are lots of other mutations,
[00:59:36.260]and this drug treatment is ideal for the common mutation.
[00:59:40.180]So if an individual has cystic fibrosis
[00:59:41.780]because of other mutations, it may or may not work.
[00:59:43.890]That particular mutation affects the ability of the protein
[00:59:47.290]to get to the membrane,
[00:59:48.520]so one of the drugs helps it do that.
[00:59:50.680]And then another drug helps it
[00:59:52.910]be more active when it's there.
[00:59:55.520]So those are really specific to that protein,
[00:59:58.800]in fact, that mutation for that protein.
[01:00:02.290]So it's that question, one question would be
[01:00:05.850]for all of these diseases and all of these alleles,
[01:00:08.730]do we want to spend all of the money and effort,
[01:00:12.030]is that the right approach, is that an efficient approach?
[01:00:16.040]Because it took 30 years to get from discovery of that gene,
[01:00:20.110]which is super genetically simple, to a treatment.
[01:00:24.740]I don't off the top of my head
[01:00:28.570]know a lot of details about other success stories.
[01:00:33.260]But I think the answer is that it's,
[01:00:35.230]you know, that's a lot of hard work,
[01:00:37.180]and it certainly can be worth doing,
[01:00:39.460]the individuals who have cystic fibrosis,
[01:00:41.170]their lives may be transformed by this.
[01:00:43.690]But it often is a lot of hard work
[01:00:46.090]to go from sequence to treatment.
[01:00:54.119]What are you finding out about junk DNA?
[01:01:00.500]It's not junk.
[01:01:04.070]So junk DNA is a term that I said, what did I say?
[01:01:07.900]Most polymorphisms don't do anything,
[01:01:10.040]and that's because most of the bases in our genome
[01:01:13.260]don't do anything obvious.
[01:01:15.720]So in our genome, about one to 2% of those bases
[01:01:19.040]encodes genes, and the rest of it doesn't.
[01:01:23.310]And so that rest of it was referred to as junk DNA,
[01:01:26.440]so the sequences that are in between genes,
[01:01:29.340]the sequences at the ends of the chromosomes,
[01:01:31.312]in the middle of chromosomes.
[01:01:33.620]And what we're learning is that,
[01:01:36.320]yeah, they don't encode proteins,
[01:01:38.770]and maybe they're not expressed, they don't function,
[01:01:42.940]but they can have other roles, they can do other things,
[01:01:48.220]and they can be important
[01:01:52.010]in scaffolding chromosomes in the nucleus,
[01:01:56.300]they can be important in turning nearby genes on and off,
[01:02:00.190]so you know, there definitely is,
[01:02:02.730]it is true that a majority of the nucleotides
[01:02:06.240]in the human genome, if you changed any one of them
[01:02:09.310]from what it is to one of the other three bases,
[01:02:11.710]it would have no effect.
[01:02:13.920]But some of these things, there's function
[01:02:16.380]where we didn't think there was before.
[01:02:18.090]And some of these things may have function
[01:02:19.860]not as a single base, but in aggregate.
[01:02:22.780]It can make a difference if you have, you know,
[01:02:24.720]a lot of our genome is very repetitive,
[01:02:26.220]it's the same sequence again and again and again.
[01:02:28.240]And it can make a big difference if you have two copies
[01:02:30.650]of that or five copies of that or 500 copies of that.
[01:02:33.750]So I think there's a lot still to be learned about
[01:02:37.830]the function of all of these bases,
[01:02:40.550]particularly the ones that aren't involved
[01:02:42.210]in coding for proteins.
[01:02:48.001]So it's a continued challenge in science
[01:02:50.649]to communicate to those who are
[01:02:53.176]maybe on the outside of science,
[01:02:54.831]with the current trends in genetics,
[01:02:56.059]do you see that challenge becoming more complex
[01:02:58.677]or less complex or somewhere in the middle?
[01:03:01.640]Oh, I think it's becoming more complex.
[01:03:03.290]But I have seen, I mean, there's been this
[01:03:07.210]explosion in genetic counseling programs.
[01:03:11.310]So this is medical providers whose job it is
[01:03:14.700]to interpret genetic information for patients.
[01:03:18.730]And so that suggests to me that
[01:03:21.830]there is a new or a growing demand for
[01:03:26.230]that kind of communication
[01:03:27.760]explicitly in the medical community.
[01:03:31.260]So I think it's getting harder,
[01:03:34.050]and I think, I didn't talk much about this,
[01:03:36.040]I think one of the reasons it's getting harder
[01:03:37.410]is that there's a lot of people
[01:03:39.230]trying to make a quick buck on this, right?
[01:03:41.380]So you can take your ancestry results,
[01:03:43.480]so your 23andMe results, and for a small fee,
[01:03:46.970]you can send those to Nutrigenomix
[01:03:49.500]which will optimize your diet
[01:03:51.350]to your particular genetic profile.
[01:03:53.440]Or you could just eat healthy.
[01:03:55.620]You can send it to Soccer Genomics,
[01:03:58.470]and they will provide a training regimen
[01:04:02.230]specified for your genotype,
[01:04:04.070]or you could just go out there and do the sprints.
[01:04:06.850]So I think that it's not gonna get simpler,
[01:04:11.700]both because the biology is complex,
[01:04:14.320]and because there's a sucker born every minute
[01:04:17.140]and there's someone at the other end of that
[01:04:18.730]standing to make a buck.
[01:04:26.036]Thank you so much for coming, have a good night.
[01:04:28.175](audience applauding)

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Gene Machines: Navigating a World Enthralled by DNA

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments