Technology Day 2001—"Origins and Beyond: Our Place in the Cosmos"

HECHT: One of my faculty colleagues says you can tell the kind of class you're in by the response. If it's an undergraduate class, there's no response. They're still asleep. If it's a group of adults, you get that kind of shout back. If it's a group of his graduate students, they write down the words, good morning.

[LAUGHTER]

For the two or three of you who don't know me, I'm Bill Hecht. And for the last 21 years, it's been my privilege to be your executive vice president. Now, I have done this for 21 years. But four times during that time are very special to me because I'm a 1, and most of you are 6's or 1's. So for the best bunch of people, the 6's and 1's, congratulate yourselves.

[APPLAUSE]

There is a small secret which I'll share, which is not a happy thought, actually. I discovered not too many years after I took this job that statistically I was rapidly becoming an older alumnus. I tried to deny that for many years. But as you can see by the ribbon, this is our 40th. So I can't deny it much anymore.

Welcome to Saturday classes. Those of you who are youngsters don't understand that this used to happen every Saturday morning. And we went usually to lab on Saturday morning and worked very hard over experiments, which we were all convinced that Michelson and Morley lied, that Millikan had oil in his glasses, not in his oil-drop experiment. But suffice it to say, we all survived. And that's part of what MIT is about, surviving in style.

It's my great privilege this morning to introduce you to yet another spectacular Technology Day program. I must tell you a little secret. For 21 years, at the end of this program we say, how in God's name are we going to equal this year, never mind be better? And each year, the volunteers and my staff, with the great collaboration of this unusual and rare and extraordinary faculty, seem to do better.

So it's my great privilege to start the morning proceedings by introducing my good friend, Chuck Vest, President of MIT.

[APPLAUSE]

VEST: Thank you, Bill, very much. And good morning to all of you. Welcome back to Cambridge. I think that we all know that universities study and learn from the past. They have some level of engagement in the future-- in the present. And they have something vaguely to do with the future.

What really sets MIT apart from most institutions, in my view, is while certainly we respect and indeed try to learn a bit from the past, we are really, among universities, very deeply engaged in the present and, above all, really working very hard to invent the future.

And another characteristic, I think, of this institution is that we tend to be involved with big, important things and to ask big, important questions. Well, today, we are going to tune this very slightly, and we're going to take a look really into the very distant past, as well as the present and the future. And we certainly are not going to shy away from big questions.

The program for today's event was inspired by this highly characteristic painting by Paul Gauguin, tropical paradise graced by stately figures who somehow manage to look simultaneously moody and serene. I've seen those expressions around this campus from time to time.

[LAUGHTER]

But there's much more going on here than in Gauguin's customary renderings of Polynesian life, because he chose to use this panel to pose some of the eternal questions in its title. Where do we come from? What are we? Where are we going?

So you'll notice that while Gauguin asks these questions brilliantly, his painting is very careful not to actually provide the answers. It's a time-honored technique, though perhaps more typically reserved for use on mid-term exams.

[LAUGHTER]

And of course, these are questions that genuinely resist being answered in any simple or final or even definitive way. But today, we're going to try anyway, this being MIT. This morning, we'll hear from four outstanding scientists who will share insights from their own lifelong pursuit of great answers about origins, the origins of our universe, our solar system, and, in a sense, of ourselves.

This afternoon, three separate panels will give us a chance to look down the road toward some thrilling new frontiers-- what we are learning about the development of mind, language, vision, and understanding, some startling new materials and technologies that will transform our lives, and the latest reports from the realm of exploration from the depths of the ocean to outer space.

Gauguin, of course, was not the only artist to consider fundamental questions of our existence. For the class that just graduated yesterday, the class of 2001, inevitably Stanley Kubrick's film 2001 has been sort of a mascot over the last four years. And in fact, that film has become such a cultural icon that it's hard to remember that it itself was based on Arthur C. Clarke's wonderful book.

But we've kind of forgotten what it was really all about. Because for all of its billing as a space odyssey, I think the deepest lesson is that the simple truth that some of the most important, undiscovered territories are to be found within us in the mysterious recesses of our own hearts and minds.

This morning, we're going to examine both ends of this spectrum of inquiry, from the sources of our universe and our planets, to the origins of humankind, right down to the level of our own genome. These are exciting times for discovery. And today promises to reflect that excitement. We're going to begin by hearing from Eric Lander, followed by Claude Canizares. We will then take a brief break, following which we will hear from Maria Zuber.

Stephen Jay Gould is unfortunately not able to be with us as planned this morning, but he has asked us to let you know that he very deeply regrets not being with us. However, we are really very fortunate to be joined by one of Steven's colleagues, Charles R. Marshall. Charles will speak after Maria. Professor Marshall, we are extremely grateful for your willingness to join us on such exceedingly short notice. And just to-- since this is MIT we'll quantify short notice. He's down putting his slides in the tray right now.

[LAUGHTER]

Following Professor Marshall's remarks, we'll have a period for question and answer back and forth between the audience and our speakers. Let me begin by introducing the first two speakers, those who will speak prior to the break. Eric Lander is professor of biology here at MIT and a member of the Whitehead Institute for Biomedical Research.

He's the director of the Whitehead/MIT Center for Genome Research, which, as you may know, was the largest single contributor to the Human Genome Project. Last February, that project published a draft sequence and initial analysis of the human genome, the set of DNA-encoded instructions that define an individual person.

Before Eric begins, I want to tell you that we have something here at MIT called Campus Preview Weekend. This is a weekend in which we invite all the young men and women who have been admitted to MIT but have not yet declared what university they will attend. And we had here in Kresge this spring a panel of MIT students so that the prospective students could ask them questions.

And one of the first questions-- one of these bright high school students held up her hand and said, can you people somehow characterize what's it like to be at MIT? What's good, what's bad? And a hand immediately shot up and said, the defining moment of my time here at MIT was when Eric Lander spoke to us all in room 10-250 about the final publication of the sequence of the human genome. That really characterized what this place is all about, and I think we would all agree with that.

Our second speaker this morning is Claude Canizares, Bruno Rossi Professor of Experimental Physics at MIT and Director of the Center for Space Research. Claude is also Associate Director of NASA's Chandra X-ray Observatory Center and leads the development of Chandra's high-energy transmission-grading spectrometer, just one of his many space astronomy missions.

While Eric is, in a sense, an expert on inner space, Claude's focus is clearly on the far reaches of outer space. He's a specialist in both X-ray and optical astronomy, meaning that he studies stars, galaxies, clusters of galaxies, quasars, and other things which give us important clues to the structure, evolution history of the universe. We begin with Eric Lander.

[APPLAUSE]

LANDER: Thanks very much, Chuck. Well, welcome to this day about origins and beyond. I'm going to start us off by talking about a variety of kinds of biological origins, origins of life, origins of human populations, origins of cancers, all of which our knowledge about emerges from the work of the Human Genome Project over the course of the last 10 years.

There's another important origin that runs throughout all of what I'm going to say, which is that the origin of an awful lot of the Human Genome Project and the data that's come out of it is here at MIT. MIT is woven completely into all of this project, both the intellectual ideas that got it started, the tremendous amount of work that got it done, and particularly the multi-disciplinary approaches involving biology, computer science, and engineering, that made it possible for it to happen here in a way that I think it couldn't have happened any place else.

So let's talk about origins. For thousands of years, humans have recognized the tremendous diversity in our own species and in the species around us. This is my favorite picture to illustrate diversity, Wilt Chamberlain and Willie Shoemaker here, showing the wonderful diversity in our own species with respect to height and weight and skin color.

And to me, it's also emblematic of the differences that you don't see, in differences of susceptibility to diabetes and hypertension and cancer, all of which are underlain by the action of genes working together with environment.

But of course, for thousands and thousands of years, we could do little more than say that traits tended to transmit in families, and we really didn't understand the origin of this transmission. Well, the major breakthrough, historically, was the work of Gregor Mendel in 1865 when he published his one great classic paper indicating that there were these abstract particles that he could infer called genes that were transmitted according to specific laws of inheritance, in crosses.

Now, as you may or may not know, Gregor Mendel's paper is a wonderful paper, but it sank like a stone. It had no impact on the world at all because the world just wasn't ready for it. Mendel, perhaps somewhat discouraged, really never wrote much more about genetics and, in fact, went into administration. And that's pretty much the last--

[LAUGHTER]

He became the abbot of the monastery.

[LAUGHTER]

In any case, the real revolution of genetics dates from about 35 years later, in the opening weeks of the 20th century. In January of the year 1900, three independent groups around the world rediscovered Mendel's laws, so to speak. They went back, and they started doing crosses. They began to see these regularities. And then they pulled out this paper and recognized what its tremendous importance was. And so the 20th century started with a bang with the understanding that there were laws that somehow controlled traits.

The first quarter of the 20th century was spent pinning down those laws to a cellular structure. By about 1925, we had a pretty good handle that the cellular origin of these abstract laws were in the chromosomes. Of course, we didn't have a clue what the chromosomes did.

Chromosome, just the word "chromosome" itself is a tribute to the scientific ability to cover up ignorance with a fancy name. The only thing that was really known about the structure of chromosomes then was that they picked up a dye when you stained the cells. And so they were called chromosomes, meaning colored things.

[LAUGHTER]

It took another 25 years to work out what these colored things really were. And by mid-century, we were able to determine that the chromosomes had, as their molecular basis, the DNA double helix, the work of Watson and Crick, this iconic double-helical image with information running on one strand and a redundant copy on the other strand. Of course, this didn't tell us how, in an information theory sense, heredity was encoded in this polymer.

But the next 25 years was spent working out the genetic code, the understanding of how the sequence of bases encoded messages, which went off to little factories in the cell that turned them into proteins through a process of translation according to a genetic code, as well as the tools of recombinant DNA, including very prominently here in Cambridge. And these tools of recombinant DNA let us do what the cell does, that is to say, copy and read out particular pieces of DNA, and even to modify those pieces.

So we got to about 1975 now with an ability to read some DNA sequence. Well, the last quarter of a century has been a time of voracious appetite to read everything we can get our hands on. It started with the reading of individual genes here and there, then more and more genes, then small genomes, the entire collections of genes in an organism, to somewhat larger genomes, to, by the end of the 20th century, a first-pass reading of the entire genome of the human being at 3 billion letters.

It's not bad for a century, to have gone from a recognition of abstract laws to essentially a comprehensive read-out of the 3 billion letters underlying that. It's a pretty good century, although I think we ain't seen nothing yet.

In the last 10 years, 10 or 15 years of that century, it began to become clear that we could use this kind of information about genetic sequence to take a systematic approach to understanding disease. Now, all through the 20th century, people worked on disease, and in some cases, they were very lucky and were able to get to the basis of a disease.

But as you well know, luck is something you can't count on any particular day when you go into the lab. Far better would be to have a systematic approach that you could apply to every disease regardless of its type. Whether it was brain degeneration or bowel irritation, you would like to have a common approach.

Well, that approach was first laid out here at MIT in about 1980. The notion was a very simple notion, that if you traced the inheritance pattern of a disease in a family, you could see that the disease appeared to be transmitted according to Mendel's laws. And if you could just find a piece of DNA which had a spelling difference that seemed to have the same pattern of transmission, then, although that piece of DNA might not be the cause of the disease, well, it must lie fairly close by.

And that is indeed this notion of genetic mapping that was laid out for the human being around 1980 or so and became operational here in Cambridge in the mid-1980s. And it led to the identification of many, many particular disease genes. Once you've found a marker that was nearby, of course, you don't have the gene itself, what you have is a long work ahead of you.

In fact, you have to go through a very tedious process called chromosome walking, or more formally, although we don't call it in the literature, chromosomal schlepping, where you would start with the marker, and then you would schlep along the chromosome, because it's incredibly tedious using one piece of DNA to get the next piece, the next piece, the next piece, the next piece, the next piece. And then four years later and $40 million later, you might arrive at the gene of interest, say cystic fibrosis on human chromosome number 7.

Well, here's the sequence of the cystic fibrosis gene. It's got a lot of letters. It doesn't look very impressive. But I've boxed for you one particular little feature over there-- I better blow that up. There we go. Those three letters CTT, those three letters are deleted in the majority of chromosomes with cystic fibrosis. About 2%, 3% of the population will carry that particular deletion.

And in fact, on the way out the door, if we wanted to, everyone could spit in a test tube, and enough buccal cells would come off from the inside of your cheek to have enough DNA that we could just take it back across the street to the lab, amplify it, and tell you if you were a carrier for that deletion. You don't have cystic fibrosis unless you have a double dose, but we could tell you if you were a heterozygous carrier.

Well, not just that, not just DNA diagnostics emerge from that, but even more emerges. One can toss the sequence of the cystic fibrosis gene into a computer and ask the computer, ever seen anything like this before? And the computer says, oh, yeah, this looks an awful lot like dozens and dozens of genes whose products sit at the surface of the cell and transport things back and forth. Congratulations. You've just found a transporter.

Now this, of course, stood on its head the usual way of doing biology. The usual way is when you make a discovery, you go back to the lab, and you spend years and years trying to figure out what this gene does. Now you spend a couple of seconds hitting the Return button, and the computer, more often than not, comes back to you and tells you what your gene does because somebody somewhere else in the world in some other organism has studied something sufficiently similar that you have a pretty good guess.

That kind of hypothesis generation by connecting all of biological knowledge has led to a dramatic explosion of research. Because in the old days, all bits of biological research were separate from each other. Now they've all been interconnected through this common lingua franca of sequence.

Well, the big problem with all this was that it was extremely tedious. It took four years, $40 million, about 200 people to get the cystic fibrosis gene out. And as we all know, the productive unit of science is the single, smart graduate student or postdoc. And they're not going to sign up for something like this. This is not a good project.

And so the job was to lower the activation barrier, to level the barriers to actually doing this, to create the infrastructure so that a single graduate student or postdoc would be limited only by her or his ideas, not by the huge amount of work that was needed. But that meant we had to engage in big science work to build infrastructure in the service of that graduate student and postdoc.

And so it was born, the Human Genome Project, the idea that we would build genetic maps of polymorphic markers up and down the chromosomes that would let us trace inheritance, physical maps of the overlapping pieces of DNA so we wouldn't have to engage in chromosome schlepping, but we could just go to the shelf and pull down the relevant piece of DNA.

Sequence maps, so we wouldn't spend years sequencing out the DNA from a region to find the gene, but we would just double click on the region. And annotations of all those genes and a description of all those genes so that when one found a gene, one would have a pretty good guess of what it did. At the time, we expected to have about 100,000 genes or so. As I'll tell you in a moment, it appears that we have rather less.

Well, more poetically, what the project was about was building for biology what the chemists had in the course of the 20th century-- a periodic table. The chemists had a periodic table from about 1869 to 1889. Following Mendeleev's key insight, the chemists put together a periodic table that described all of matter in terms of about 100 building blocks.

Those building blocks were understood. Their relationships were summarized in these rows and columns of the periodic table, giving rise to the chemical industry, giving rise to quantum mechanics to explain all this. The variants on these elements-- the isotopes were understood. And this is, of course, the basis of all of chemistry.

Biology is now at the point where it's putting together its periodic table. There are just a finite number of genes. May sound like a big number, tens of thousands, 30,000 genes. But that's a pretty small number these days. We will, in relatively short order, know them all in pretty good detail, know all their variation, know what they do. And the biology student a decade or so from now will start with a pretty good periodic table of all the components.

Now, how is all this done? Well, from about 1985 when the Human Genome Project was first conceived of, in 1990 when it was actually launched, we needed to make just dramatic changes in the way biology was done. Traditionally, the way you would sequence DNA was you would clone the DNA in a little piece in a bacterial cell. You would grow up the bacterial cell on a Petri plate.

You'd take a toothpick, you'd pick it. You'd inoculate a culture. It would grow overnight till you had enough of that bacterial colony with that one piece of DNA. You'd centrifuge it down. You'd use organic reagents to crack it-- or you use reagents to crack it open and organic solvents to be able to separate the protein from the DNA. Then you'd, et cetera-- oh, god, it was just terrible and really slow. And it was done one at a time with pipettes.

So a big concern in the period 1985 to 1990 about the Human Genome Project was, would we consign an entire generation of graduate students to some kind of servitude--

[LAUGHTER]

--sequencing DNA? Was this going to be deadly boring? There was talk about doing this in criminal institute-- in penal facilities.

[LAUGHTER]

Sentencing people to 10 megabases--

[LAUGHTER]

--with time off for accuracy.

[LAUGHTER]

Well, it didn't turn out that way. And it didn't turn out that way because biology met engineering. And so just a couple of blocks away from here, here at MIT's Genome Center, this is the equivalent of the toothpick. The toothpick now is represented by robotics that run around the clock. They have cameras on them that identify where the colonies are and 96 pins that go and pick the colonies, dump them into media, wash themselves, bake themselves, go pick 96 more, et cetera.

And we pick 110,000 distinct bacterial colonies a day. And all that a person has to do is put the plates on and walk away. We don't have to purify all of these DNAs, set up sequencing reactions, et cetera. Instead of the army of people standing around with pipettes, it looks like this. This is the way we do mini-preps for DNA. And this is very much unlike a biology lab 10 years ago.

What you see are factory-style conveyor belts with microtiter plates moving around the laboratory at a pulse rate of one per minute. 110,000 of these items being processed through the laboratory, all the way from DNA purification through the set-up and clean-up of a sequencing reaction.

Once all of that biochemistry is done, we have to put them through detectors. There, the commercial world has served us well by developing detector machines that we can use. We have a farm of these sequence detectors. We put on 10 plates on each sequence detector, walk away, and it runs 24 hours unattended. All told, this whole operation I've described to you, the actual number of people involved in managing the workflow there is under about a dozen and a half people managing all of that workflow.

At the end of the day, this suite of machines turn out about 65 million letters of DNA sequence per day. This is just a tremendous explosion of information. At the height of the Human Genome Project, the rate of sequencing was 1,000 letters of new sequence every second.

As best I can tell-- I don't have the actual numbers-- but as best I can estimate, I think that rate has actually doubled since then. Despite the fact that the Human Genome Project is no longer in the height of the frenzy, so much capacity has been in place that I think the rate is continuing to go up.

Well, all of this led to a dramatic explosion. The sequencing of the human genome was first piloted from about 1996 to 1999. The pilot period was when we worked out all this robotics, all the biological technologies, to be sure that we could cover the sequence of genomes, et cetera.

We then had scheduled, and it had been scheduled for quite some time, a ramp-up in the spring of '99 to try to take on the sequencing of the genome. And lo and behold, everything was ready for it. The ramp-up was launched. And in the course of about 12 months, we went from having about 10% of the genome covered to having nearly 90 of the human genome covered.

It looked something like this. That was the sequencing of the human genome over the course of that month. And it was just-- I mean, I just love looking at that picture there. We'll do that again. That was just so much fun. There we go.

[LAUGHTER]

How is the genome getting filled in? I don't know. To me, there was a lot of pain underlying that picture. But sort of like childbirth or something, in retrospect it's really just fun to think about there. In any case, the product justifies it. This was the product, not just of MIT, of course. It was the product of an international consortium of 20 different groups around the world.

Indeed, MIT was the largest contributor to this effort. We contributed something in the neighborhood of about 30% of the total sequence, and it really was quite a big effort here. But there were also large centers in England, the Sanger Center in Cambridge, England, and also at Washington University in St. Louis.

But I think it was equally important that there were laboratories in many other countries of many different sizes that contributed to this-- the United States, in the UK, in France, in Germany, in Japan, and even in China. The group in China joined in late, but that group in Beijing did a marvelous job. They contributed 1 and 1/2% of the sequence of the human.

And what it said clearly was that the human genome was indeed the patrimony of all of humankind and appropriately was worked out by a large international consortium. The only criterion for signing up for the Human Genome Project? You had to agree to make all of your data immediately and freely available on the web as soon as it was assembled. Within 24 hours of assembling the data, it went up on the web, no restrictions whatsoever.

All the groups signed up for it. And I think it was a wonderful thing that they did because it meant that people could immediately start using this information. When we went to write a paper about the sequence of the human genome, well, usually people have to speculate in their paper about what the consequences of their work will be. We had the absolute joy of being able to devote a section of our paper to how our data had already been used to find disease genes. And that was just a very satisfying aspect to it.

So anyway, this whole Human Genome Project got a fair amount of attention. There was lots of hoopla in the press, and there were several visits down to Washington, some celebrations down at the White House and things. But to the scientific community, the only thing that really matters is a peer-reviewed scientific paper. And that occurred in February with the publication of the largest paper in the history of the journal Nature on the initial sequencing and analysis of the human genome.

It actually was written as a friendly paper. And even if you're not a biologist or haven't touched biology in a long time, you might want to pick up the paper because there are whole chunks of it that really were intended to be friendly enough to be readable. There are other bits you'll want to skip over, but they'll be obvious. So just skip them over.

I'm not going to try to tell you everything that's in this human genome, but I'll mention a few interesting things, particularly with regard to origins. Interpreting the sequencing of the genome. We have a fold-out to the sequence of the human genome. It's actually two entire centerfolds on both sides.

This is just chromosome number 11 here that I'm demonstrating, and you'll see the gene distribution and many other features. You can't really see the details. But I'll call your attention to the fact that there are some gene-rich regions and some gene-poor regions-- gene-rich, gene-poor, gene-poor, gene-rich. And interestingly, this correlates with the visible banding patterns of the chromosome. This is part of a dark band on the chromosome. This is a light band on the chromosome-- dark band, light band.

So even sitting way back there in the bleachers, you can see that there's a correlation between banding pattern and gene density. And if your eyes really good, you'll even see that there's a correlation between low gene density and a low proportion of G's and C's in the sequence as compared to A's and T's.

Anyway, there are many gross interesting features that are visible at this high level of view. Of course, this is not a very good view to take because, well, at least as it appears in the journal, the scale is about 3 million DNA letters per centimeter, which you have to really good visual acuity to appreciate.

[LAUGHTER]

And so instead, there are a number of genome browsers, freely-available genome browsers that you can just find on the web. You can look to the University of California Santa Cruz, the National Center for Biotechnology Information, to the European Bioinformatics Institute.

Lots of cooperating genome browsers are there, where you can zoom in by factors of 10 all the way down to the base and up all the way to the chromosome. You can scroll over. You can see whatever annotations you want or hide those you don't want want. And it's a wonderful thing that we now have a number of these freely-available browsers that let you inspect the genome.

I'll mention just one or two facts about the genome that are kind of fun. Most of your genome does not consist of genes. In fact, genes are only about 1 and 1/2% of your genome. Most of your genome consists of repetitive elements that are autonomous, replicating elements that copy themselves, and on evolutionary time scales, move around your genome. They hop around.

So if this is your genome here, half of it is filled with these repetitive elements who are interested only in their own reproduction. They make copies of themselves into RNA. Those copies are copied back into DNA, and they get slammed into the chromosome. See? There we go. They get pushed right into the chromosome there. And over the course of time, they account for about half of your total sequence.

Now, what's very interesting about these repeat elements is whenever they hop, whenever they do this-- they go womp, womp-- they bring with it the sequence of the ancestral elements from which the copy was made. But when they land there, they're usually broken, and so they begin to build up mutations. That means they're a historical relic, and we can tell how old they are by how much they've built up mutations.

In fact, we can tell whether two copies of this repeated element are cousins or close brothers that were born at the same time because they have the same distinctive sequence characteristics. In fact, we can build an entire phylogenetic tree relating the origins of all of these repeated elements in your genome to tell which ones are part of a sibship that hopped a million years ago, or dispersed themselves 20 million years ago, or as far back as 800 million years ago.

All of that is there. So our chromosomes, we now realize, are a geological record. There are all sorts of geological fossils, about 3 million geological fossils, all of which have dates on them.

From that, we can, for example, tell, just the simplest thing, what the rate of this transposition has been over the course of time. If we go all the way back, reaching back beyond half a billion years, about 3/4 of a billion years, the rate was sort of constant. It shot up around 75 million years ago.

And then, interestingly, the rate at which these elements have been hopping around our genome has plummeted in the last 30 million years. This is a major ecological change. We don't understand what's happened. Why have these things stopped hopping?

This, by the way, has consequences. Since this accounts for half your genome, this means that in the long run our genomes are going to get smaller because there's a deletion process that deletes these. And if you're not replacing them, the equilibrium size of the genome's going to get smaller.

Now, don't worry too much about it because the time constant on this is pretty long. The rate of deletion of useless DNA is about-- well, we houseclean our useless DNA with a half-life of about 800 million years. So this effect will kick in in the next several billion years or so, but nothing to get too excited about.

But it is very interesting biologically because this fact I'm telling you is not true about the mouse. In the mouse, it appears that the rate of transpositions continue just fine. It's something different between the hominid lineage and the rodent lineage. And if I had time, I'd tell you about a population genetic hypothesis that explains it in terms of the effective population size of hominids versus rodents. But we're understanding now, first of all, the origin of half of our genome and a fascinating dynamic process of these elements.

Well, there's much else you can tell about the genome by following these passive evolutionary markers that are just conveniently scattered across. But I won't tell you about that. I'll describe briefly the genes. There's a lot we can tell about origins from the genes themselves.

There was a great effort in this paper to try to do the best job we could of gene identification. It's by no means absolutely complete, but I think the vast majority have been identified. And the first surprise was there were a lot fewer genes than we had thought.

Indeed, I have apologies due to some members of the audience. I have been lying to the 701 students, the Intro Bio students, for the past 10 years by teaching them that there are 100,000 genes in the human genome. I'm trying to send out notes to all of them and make sure that--

[LAUGHTER]

--we correct this. But it appears that there really are only 35,000 genes in the genome. The 100,000 number, when we closely checked the literature, a, it turns out to be a back-of-the-envelope estimate that had really huge error bars that somehow everybody just latched onto, and b, in any case, the 100,000 estimate was due to a colleague at Harvard.

[LAUGHTER]

So the genes tend to-- these genes are very interesting. They bunch up in these GC-rich regions. They have more splicing. We have more alternative splicing, alternative ways to splice and dice our genes and gene transcripts, than happen in flies and worms and things.

But when we actually look at the origins of humans, we should ask, are our proteins very different than other organisms in terms of the modules they're made up of? And the answer is no. Maybe 5% or so of the protein modules, the little architectural domains that make up our proteins, are innovations amongst vertebrates. 95% of them were there before vertebrates.

So what's new about us? Well, the origin of vertebrates seems to have more to do with innovation in architectures rather than in the domains themselves. By architectures, I mean ways of combining the domains in new combinations but existing domains. So in point of fact, our creativity that gave rise to the vertebrates is somewhat derivative. It's mixing and matching of pre-existing parts.

In a sense, it's not that surprising, of course, because these pre-existing parts-- back in the primordial ooze, you had time to evolve a new domain, and it didn't have to work so well. But as life filled up more and more of the niches on this planet, it was necessary in order to compete that a new gene pretty much work from the get go, and therefore had to be made out of pre-existing components. And we can see that very clearly in our DNA that that seems to be the basis of vertebrate novelty.

In addition, the simplest way to make novel genes is just to duplicate a pre-existing gene and let it change slightly. Highly derivative, but very effective. And this has been the basis, for example, for such things as smell receptors. You have in your genome 1,000 smell receptors, indicating that vertebrates were very interested in smell. The origin of vertebrates has a lot to do with smell.

But then, unfortunately, I have to tell you that you have largely decided that, despite your ancestors' interest in smell, that you are not so interested in smell. Hominids have largely let these smell receptors go to seed. About 2/3 of your smell receptor genes are broken. They have stop codons. They no longer work.

That is not true about a mouse. Mice are much more interested in smell. They keep their smell receptor genes in fine working order. We seem to have put all this energy into making smell receptors, and then we let them break in favor of sight as a preferred sense for humans. All of this you can tell from the genome. There are stories about origins and changes throughout the genome.

Well, anyway. Whoops, back. I can't tell you all of the wonderful things in the sequence of the genome. There is so much to say about what's in there. But what I can tell you is that it's just the start. We went into the Human Genome Project thinking that we had to build infrastructure in the form of maps and sequences in order to let people really do their work, in order so that people would be limited only by their ideas.

We've come to understand that the work of building infrastructure has only just begun. There is a tremendous amount of work still to go. I want to just highlight some of the areas in which that's going on. Comparative genomics. To really understand the sequence of the human, to understand medicine, to understand origins, we need to be able to compare the human sequence with many other organisms to see what evolution has kept and what it's changed.

We need to have comprehensive descriptions of each gene in terms of where it's expressed in the body and when it's expressed in health and in disease and during development. We need to know the shape of every one of the proteins, so-called structural genomics. We need to have a chemical that interacts with and binds to and inhibits, where possible, every protein.

We need to have diseases classified in terms of the sets of genes that are turned on and off. And we need to have a comprehensive description of the variation in the human population. All of these are giving rise to genome-style projects of their own to lay the groundwork for that graduate student and postdoc of the future.

I'll just touch very briefly on where all this is going just to give you a sense of the whirlwind that is starting. If you thought the Human Genome Project was something, just wait for the next decade of what's going to explode. The mouse. A year ago, we had very little sequence of the mouse, maybe about 1%.

As of today, about threefold random coverage of the mouse has already been generated. It covers about 90% of the entire genome. It's freely available on the web. And it's been integrated with the human sequence by matching them up. In fact, 2/3 of this entire project was done here at MIT, and it is leading to a tremendous amount of work on mammalian genomics.

It's important because by lining up mouse and human, we can see the conservation of elements, and we can spot things that we might have missed with our computer program. For example, if we ran a computer program over the human genome to identify genes, we have a signal-to-noise problem. Genes are only a little bit of the overall sequence.

By lining up the human and the mouse together and running a computer program over both simultaneously, we can kind of do a kind of coincidence detection, almost a genomic interferometry of some sort, where we ask, do we see features present in both genomes? And we get tremendously higher sensitivity and specificity to identify genes.

And not just genes, but what's popping out of things that we had never had a way to see before. Regulatory elements that don't encode proteins, but encode the instructions to turn on here at a certain time we recognize because of conservation between mouse and human.

There is, of course, work on many other species. In the course of the past year, we sequenced a very important and largest fungus studied, Neurospora, a very important genetic organism. And this was just a great example. We got a three-year grant from the National Science Foundation to sequence Neurospora, and we completed 99 and 1/2% of the Neurospora genome in the first two weeks of that grant. The last half a percent will take a while, I admit, and there will be time spent analyzing it.

But the success of this project here at MIT has led to a consortium of fungal geneticists to identify 15 different organisms that they want to study, representing important medical fungi, commercial fungi, and fungi that will shed light on evolution and diversity. And so we are hoping, if we can get the funding to do this, to basically start a Fungus of the Month Club starting later this year.

[LAUGHTER]

A model vertebrate, Tetraodon, the pufferfish. The sequence of this pufferfish is being generated in Paris and here at MIT. And already, about threefold coverage of this organism is available. This seemingly unimpressive filter feeder, a tunicate called Ciona, it just sits there and filter feeds.

Why bother sequencing it? Well, it turns out it has a brief period in its life when it is a tadpole, and it has a complete chordate body plan with a notochord and all the structures in a tiny genome. And so we began sequencing this beast a few weeks ago. We've already got onefold coverage of its genome, and we expect to have about tenfold coverage of its genome this year.

Onward and onward. Genome sequences will come spewing out, similarly, comparisons amongst human beings. There is a tremendous amount of sequence in the human genome. This is sequence. But there's not that much variation. That's the variation. If you can't see that, I'll circle that for you. There we go. Those were the two spots of variation. That's it. One base in 1,300 is the extent of variation in the genome.

It turns out that this makes perfect sense from some dusty, old formulas of population genetics. Little populations have a little bit of variation. We're actually a very little population. Although we're 6 billion people today, we were 10,000 people in Africa a mere several thousand generations ago. And that signature of a little bit of variation a few thousand generations ago has persisted because a few thousand generations isn't enough time to build up more variation.

So as a consequence of that, we predict that most genes should only have two or three flavors in the population. And when we look, that's exactly right. Most genes have only a few flavors. That has led to a very exciting idea that every gene has its isotopes, a few isotopes, and that we could correlate disease with the variants, or the isotopes.

This common disease/common variant notion is, I think, becoming very important. We know some examples-- apolipoprotein E. It comes in three flavors, lives on chromosome 19, E2, E3, E4. If you have a double dose of E4, you have a high lifetime risk of Alzheimer's disease.

We know a few other examples like that. But our goal is to identify each and every one of the variants for each and every one of the genes, that is the common variants, and correlate them with all major traits. And while this seemed like an absolutely nutty idea even five years ago, I think folks are conceding that this is very likely to happen certainly over the next decade. And it may be that we have all this variation characterized much sooner than that.

Indeed, the goal was to detect disease genes using lots of variants. There are two different ways to do it. I won't fuss over them. We started with a paper here at MIT about three years ago, where we reported 4,000 variants in the human genome. We thought that was a lot at the time. Of course, it now is a puny number.

But what it did was that particular paper got the world interested in generating these variants. And an international consortium was put together in parallel with the human, and it led to the generation of 1.42 million variants across the human genome. And this paper was published in parallel with the human sequence paper back in February. This, again, was a wonderful international consortium. And again, it was a consortium led here at MIT with about half of all of this work having come from MIT.

With it, we are now studying the nature of these ancestral segments in populations, and it's really quite a remarkable thing. The segments that we're looking for, they are big. They have a limited number of forms. And I believe we are going to be able to get comprehensive descriptions of all of the ancestral types of all of the genes. And I think that this will form the basis of a new kind of human genetics.