More Science Talk
May 17, 2006 -- Inside the Tevatron; the Human-Computer Interface; DNA Computing.
Male voice: Novartis—committed to making innovative medicines for a world of patients and their families, online at novartis.com Novartis…. Think what's possible.
Steve: Welcome to Science Talk, the podcast of Scientific American for the seven days starting May 17th. I am Steve Mirsky. On this week's podcast, we're going underground with Scientific American magazine editor Mark Alpert, talking about his visit to the Fermi National Accelerator Laboratory. Then we'll be talking about computing, dry and wet. Scientific American senior writer, Wayt Gibbs reports from a conference devoted to the interface between computers and humans and Ehud Shapiro from the Weizmann Institute of Science talks about computers in humans—DNA computers.
First up, Mark Alpert. He is one of SciAm's physics guys; as you[']ll hear, they think climbing inside a particle accelerator beats a trip to Disney world. I spoke to him at his office.
Steve: Mark you, were telling me you had an interesting experience last week. Where were you?
Alpert: Well I was at Fermi National Accelerator Laboratory inside the Tevatron, which is a pretty unique thing to do, because usually you can't go inside there because the beam is running and there'll be some radioactivity in the tunnel, and so they don't let people in. But the Tevatron was shut down for ….
Steve: Well back up a bit. Tell everybody—what is the Tevatron and what were you doing there?
Alpert: Well the Tevatron is the still world's most powerful particle accelerator. Basically it accelerates protons to super high speeds 99.9999 et cetera percent of the speed of light; and then it smashes these protons against some antiprotons and physicists study those collisions in the detectors to see what comes out of it. And that explains things like you know the nature of matter, the existence of other dimensions, things like that. And the Tevatron has been around for more than 20 years but it's going to be replaced pretty soon by the LHC, the Large Hadron Collider in Europe. So these are the last years of the Tevatron's life, but it's still producing some really interesting physics. They are still working at it trying to push as many protons as possible into that beam, so that they could discover more things in these last few years of the Tevatron's life. So I thought, I need to go there to see it before it shut down; and luckily the time I went the Tevatron was in a shutdown mode—and that means they were doing maintenance on the thing—so the protons weren't actually running, so I could actually go inside the tunnel. And now this is a tunnel that's about 25 or 30 feet beneath the surface of the earth and it curves in a big ring 4 miles in circumference; and it's in the middle of Fermi National Laboratory which is this, you know, beautiful stretch of preserved prairie, right outside of Chicago.
Steve: So what did you get out of it other than the experience of being in there or anything?
Alpert: Well, just, I was amazed at the expertise of these people who run the thing—about how they had used all these little tricks to get it to work at maximum efficiency. I mean, the thing about the Tevatron is that they can't increase the total energy to collision. You have protons moving at one trillion electron volts in one direction, antiprotons moving in the other direction in one trillion electrons volts, so [a] total energy of two trillion electron volts. You can't increase the energy, but what you could really increase is, like I said, that efficiency—the number of protons hitting each other—and apparently they've done so many little tricks on the system that the temperature on it is working at an efficiency of two hundred times higher than it was designed for.
Steve: That's pretty impressive.
Alpert: Yeah! And it was just really fun, I mean you know, just seeing the people and going into the control room and they have this, you know, these computer screens, and there is this big zipper running across the top saying, Tevatron is in shutdown mode; and everyone calls it the Tev! You know this is their, you know, buddy or something.
Steve: And that goes off line when the Large Hadron goes online.
Alpert: Well that hasn't been established yet; I mean, there is still a lot more the Tevatron can do. I think the scheduled date is something on the order of 2010, so you can run it for a few years. It all depends on what happens with the Large Hadron Collider, which I think is supposed to come online in 2007, but may not be fully online. You have to calibrate it, and the full results coming in, but eventually the high-energy frontier will move to Europe; but there is a lot of exciting things that are still going to happen at Fermilab. They are now reconfiguring a lot of the lab so that some of their assets can be used for research into neutrinos. They've already got a nice neutrino program at Fermilab. They got this one experiment called Boon, where the[y] fire a neutrino beam at this huge spherical tank that contains 800 tons of mineral oil—and neutrinos, as you may know, are particles, very mysterious particles—they have no charge. Neutrino means little neutral one, and until 1998, people thought they had no mass either; but then they discovered that these things are weirder than anyone thought—they are actually changing flavor as they fly through space, and that indicates that they do have mass. And so physicists were suddenly stunned by this, and now they are really intensively investigating how they're changing their so-called flavor from electron neutrinos to muon neutrinos to tau neutrinos. And one of these experiments is this Boon experiment where they fire muon neutrinos at this spherical tank of mineral oil and they see how many of them have changed en route.
Steve: David Blaine is not in this spherical tank for the mineral oil.
Alpert: (laughs) No, no,
t the tank has been shut since 2002, so if he is in there, he's in trouble. It's basically [a] closed up tank, but on the inside surface of the sphere, are these photo multipliers; and so what happens is when one of these neutrinos comes in, the neutrinos— t[w]hat will make some so devilish is that they interact so infrequently with anything, I mean there is trillions passing through you all the time—but every once in a while, they'll hit something, and they'll do something. And hopefully when you fire an intense neutrino beam at this tank, every once in a while you see this little flash, little scintillation come out, and it's this thing called the Cherenkov light: It's a cone of light and they could tell from the characteristics of this light, you know, what was that—was that an electron neutrino, was than [that] a muon neutrino, or was that, you know, something else? You know, and so they analyze the result and they've been analyzing and analyzing and analyzing; and supposedly sometime this summer, they are supposed to come out with their big result. And this result could be really kind of revolutionary because what they're trying to confirm is that there might be a fourth kind of neutrino, which some people call the sterile neutrino which interacts even less than the flavored neutrinos.
Steve: Now the whole deal on neutrinos, I think, tell me if this is right is that little bit of mass that they look like they have is really important for everybody who's trying to figure out—what the heck is going on in our universe here
with[if] there is a lot of mass and energy that's unaccounted for?—and there's been hope that they could blame it on the neutrinos, is that right?
Alpert: Well yeah! But with the clarification [that] there are certain upper bounds that they've established [for] a neutrino mass. So most theoretician[s] seem to think there is probably no more than—and if neutrinos can probably account for—no more than a few percent of what they call the dark matter, the missing mass out there. But there is another thing called dark energy, which is even more mysterious which no one really can get a hand on yet. Possibly our neutrinos are involved in that—who knows? And what makes neutrinos so exciting is people understand them so poorly; and the physicists I talked to said, Look, if you're looking for physics that we don't know yet, physics beyond the standard model, new kinds of physics, the neutrino is a good place to look,
for because we don't know what they're doing—maybe some new kind of physics; maybe what we're seeing with neutrinos is some manifestation of the new physics, and that's what lot of these experiments are viewed at finding out.
Steve: One of the things I really like about science and scientists is that they are—out in the rest of the world, when things are not understood, people sometimes get aggravated and frustrated, and that happens in science too. But you hear a lot of scientists talk with such enthusiasm and excitement when things are not understood because that's the next place to go to explore.
Alpert: Yeah! I think that's true. You sense
the[that] scientists' eyes are lighting up when they bring up concepts that sound incredibly radical, and, of course, you know, they say with any radical discovery [it] requires [a] really extraordinary approach—extraordinary discovery requires extraordinary approach—and so that's why you know, they are so intent on confirming experiments. You need at least two experiments to really confirm something about particle physics. And what's interesting to me too is when you talk to the experimental guys, the guys who are designing and running these big detectors, their whole mindset is very different from the theoretical guys. And when talking about the theory, I mean, they are very excited about getting a new result and they really want to make sure it's a good result they want. They really bend over backwards to do their analysis of the results correctly to make sure that they are not just making something up out of the computer data, that the result is real. But then when you ask them, well, what does this mean? they sort of hand off the ball to the theoreticians; they say, Oh! … yeah! … well, it says a lot about the oscillations, and then I start asking them even more about that and they say, yeah you're getting a little behind[beyond] my expertise even, you should talk to a theorist. So it's really neat to see the different kinds of outlooks that two kinds of scientists have.
Steve: Great. Thanks a lot Mark.
Alpert: Oh! You're welcome.
Steve: The Web site of Fermi National Accelerator Laboratory is www.fnal.gov. We'll be right back.
(Robotic voice: Greetings human. When there is some thoughts about the podcast, let us know what you think by participating in our survey at www.sciam.com/research.)
Steve: Now it's time to play TOTALL…….Y BOGUS. Here are four science stories, but only three are true. See if you know which story is TOTALL…….Y BOGUS.
Story number 1: A quick and easy dip strip could soon be available for testing hot beverages for caffeine, thanks to antibodies from camels and llamas.
Story number 2: It's long been assumed that for effective hormonal signaling, you want a tight fit between a hormone and its receptor, but a new study show that a loose fit between a hormone molecule and its receptor may actually ramp up the level of cellular signaling that the hormone is there for.
Story number 3: A bill being considered in England would ban ice cream trucks from the vicinity of schools because the constant repetition of the ice cream truck song causes children to become distracted and hyperactive.
And Story number 4: A study shows that microbes can create usable electricity as they breakdown organic material in wastewater.
We'll be back with the answer. But first, Scientific American senior writer Wayt Gibbs was in Montreal in late April at the CHI 2006 Conference. CHI is for computer-human interface, and the conference is the largest annual meeting of computer scientists who study and invent the ways that humans and computers talk to each other. Wayt sat down with Ed Cutrell, a human, and a member of Microsoft Research's Adaptive Systems Interaction Group. Here's Wayt's report.
Gibbs: Cutrell's group at Microsoft Research has been developing software called Phlat that goes a step or two beyond
the Google for the desktop utilities that are available now, that allow people to find information on their own computers rather than the Internet using a Google-like keyword search. Phlat gives users the ability to tag their documents, photos, e-mail messages and the like with labels to create whatever set of categories make sense to them. They can apply these tags to any other things that show up in their search results. Cutrell argues that in the long term, this kind of non-hierarchal organization system is likely to supersede the user virtual folders as storage bands for the data on our computers.
Cutrell: One of the other things that we're really interested in [in] our group, that Susan Dumais and I've been working on for quite a while, is something called the implicit query. People don't search just because they feel like searching. Their searching, it's a part of something else. They're always looking for some information. They can do something else. It's an ongoing task and search is somehow supporting that by finding some information for them.
Gibbs: To take a trivial example, you need to find an e-mail message from somebody because you're preparing to send them an e-mail message, and they reply …
Cutrell: Exactly—you need their address. Anytime you do a search, it always happens with something else in mind or doing something else [and]
into the extent that we can, [we want to] start to understand what that is, because we are to help you out. Implicit query is one way to do that. What implicit query means is—if contrasted with explicit query, where you type in to Word or you are going out saying, I am searching for a blog—in implicit query, the machine just does something for you. It does the search, presents you the results and ideally you're either not distracted if you didn't care about it or terribly pleased if you did.
Gibbs: It's Web-distant, time information delivery.
Cutrell: Exactly. There are a bunch of challenges, yes; I mean, one is figuring out what you are doing and the other is figuring out what's relevant to you. Third is how you present this information in a way that's not obtrusive, in a way that doesn't just bother you so much you turn the whole thing off. Fourth is that if you need it, you can see it, because we just built a little system that's a plug-in to Outlook. It's like, pull up a mail message and what the system is doing is going through and it's reading all the text in the mail message and starting to look up of all that text against my index, my personal index. Conceptualize it this way: Your index—Windows desktop searcher, Google searcher, whoever—that index is basically a list of all the words
and[in] all your documents— how often they happen and where they occur. Think of that as being a proxy for you. Now I have an arbitrary e-mail message come in. I look at all those words and I compare those words in that document to the words in the index and say, you know what, these words here are not very useful because they occur all over the place in this index.
Cutrell: These other words on the other hand don't happen that often, so it can first do [a] good search function of the message, then it does that query automatically, puts it up over here, so what happens is that it looks over all the stuff and it just shows related things. So this is an e-mail message from Susan asking me to send her my slide deck from this talk again the other day; and what you see are a series of e-mail messages about various things, some of whom are about some competition and about some slides here and other stuff; and in fact you can see the slide deck which she has asked about, it's right here. There's a lot of challenges here about how to show this information. One of the challenges is the fact that every time I try to e-mail or do anything in e-mail, I am getting this thing of slide, which is showing
about[all] this stuff; that can be very distracting. And if I got lots of e-mails open, and they all have this thing, it's going to be even more of a problem, so one of the things we do is, for instance, if the mail message is moved around, or if it's anywhere else, we just hide that little window. If I have multiple e-mail messages open then only the one on the top is present. Maybe I can copy or paste something, As soon as I've highlighted it, this is now then updating according to what I just did. It's basically a new query for just those terms. So as you can see, basically as I am typing, it's just updating this as I add more and more words.
Cutrell: Now that's a little bit distracting, but we found that it's not that bad, that most of the time that this has proven to be useful is when it's presenting the stuff that I didn't know I needed to know. So what we're trying to do is provide some peripheral awareness of that information that's not too overwhelming and not too distracting. It's arguable whether we hit up with this but so far it's pretty promising.
Steve: Phlat is available for download from Microsoft Research at research.microsoft.com/adapt/phlat; and that's p-h-l-a-t. Implicit query, which was first described several years ago but has only recently entered active development, is currently being tested internally by several hundred Microsoft employees. Back to Wayt Gibbs with more from the Computer Human Interface Conference.
Gibbs: A number of other intriguing prototypes and research results were presented at the conference, including a brain-wave monitor for better spying, hi-tech toys to help toddlers talk and a study on the hazards of skimming through Scientific American. First, if you ever worry that the CIA is monitoring your brain waves, you might be interested in a work presented by a group of researchers from Honeywell Laboratories and Oregon Health and Science University. Noting that the majority of military surveillance imagery never gets examined, they turned to brain scans as a way to help spies scan more quickly through large sets of images—in particular satellite surveillance photos—and to identify those pictures that contain targets of interests, such as say missile silos or insurgent camps. The scientists outfitted professional image analysts with helmets that picked up the electrical brain waves produced as
this[these] subjects viewed photos flashed in quick succession on a computer screen. And[In] experiments with tens of thousands of images, the scientists noticed that when the analyst looked at an image that contained a target, a distinctive wave of electrical activity often swept from the front to the back of the brain. They programmed a computer to use the brain wave pattern to discriminate interesting from uninteresting images and they found that they correctly classified the images about 85 percent of the time. Parents wanting to give their infants that crucial edge in preschool will want to follow the visiBabble project. In Montreal, Harriet Fell and co-workers at Northeastern University presented hi-tech visiBabble toys that they have designed to help toddlers or older children with speech delays learn to talk. The device has monitored the babbling of youngsters and the word utterances that are similar to actual syllables. By offering encouragement such as cheering sounds or cute photos, the prototype system increased syllable production by about a half a nd syllable [and] variety by about a third in preliminary tests. Finally, the conference sounded a note of warning about trying to blow through that issue of Scientific American too quickly. Two informatics experts at the University of Manchester in England asked 32 undergraduate psychology students to read articles from Scientific American on a computer screen at either a normal speed of 225 words a minute or a skimming speed of 600 words a minute. Comprehension test revealed predictably that readers were less likely to recall particular sentences from the article when they zipped through it quickly. More surprisingly, however, the researchers found that when they asked the students to evaluate false statements about the topic, skimmers were much more likely to say that the statement was consistent with the article, which, needless to say, it wasn't. This research might explain some of the puzzling letters we get from readers from time to time.
Steve: More info about the Computer Human Interface Conference can be found at www.chi2006.org/blogs/official. We'll be right back.
Male voice: Novartis—committed to making innovative medicines for a world of patients and their families, online at novartis.com
Novartis…. Think what's possible.
Steve: Now it's time to see which story was TOTALL…….Y BOGUS. Let's review the four stories.
Story number 1: New caffeine test has camels and llamas to thank.
Story number 2: Looser fitting hormones may be what the doctor ordered.
Story number 3: Ice cream trucks' song banned from British schools
And story number 4: Electricity from microbial activity in wastewater.
Story number 1 is true. Antibodies from camels and llamas for some reason can survive at much higher temperatures than antibodies from, say, mice. Camel and llama antibodies to caffeine could thus be harvested and used in a dip strip, like for a litmus test or a blood glucose test. You dip the strip in the hot beverage and observe some kind of color change if caffeine were there. The research was in the journal Analytical Chemistry.
Story number 2 is true. A news study shows that a hormone that doesn't quite fit its receptor may actually do a better job. That’s according to a study in the current journal of biological chemistry. Researchers looked at thyrotropin releasing hormone, which stimulates the thyroid. They also examined six slightly altered versions of the hormone, some of which turned out to be twice as effective as the real thing. The thinking is that the looser fitting, slightly-wrong hormone molecule keeps kind of clicking in and out [of] the receptor, with each click in fooling the receptor into sending off its signal. Look for that concept
of figure in future drug development.
And story number 4 is true. By making fuel cells using the right kinds of bacteria as they digested organic waste, researchers got a usable electrical current. You can read more about that in David Biello's article "Microbes Convert Wastewater into Usable Electricity". It's on our Web site, www.sciam.com.
All of which means that Story number 3, about the British ice cream trucks being banned from near schools because the song drives kids nuts, is TOTALL…….Y BOGUS. What is true, however, is that a bill in England may indeed ban the country's 5,000 ice cream trucks from going near schools in an effort to fight childhood obesity. And what is also true is that the story in the Times of London began by noting that the familiar ice cream truck theme song might no longer be heard near schools, and I honestly assume[d] that the story was going to be about banning trucks because of the negative mental effects of hearing that theme music over and over and over and over … okay … okay… I'm okay. In England, it's apparently Greensleeves on the ice cream trucks, which has got to be better than what we hear in New York which is this (ice cream truck music Plays), which I actually found—somebody posted it on the Internet. That music has crawled into my head so deeply sometimes I've considered writing variations on it for bells and straightjacket flaps.
Next up, Ehud Shapiro, he's a professor in the departments of computer science and biological chemistry at the Weizmann Institute of Science in Rehovot, Israel. And he and Yaakov Benenson wrote the article in the May issue of Scientific American called "Bringing DNA Computers to Life". I called him at his home in the small village of Nataf in the Judea Mountains.
Steve: Dr. Shapiro, great to talk to you today.
Shapiro: Hello. Good afternoon.
Steve: The article in the current issue of Scientific American, very interesting, about cellular computing and biochemical computing. One of the things at the end of the article—let me start with the end. You talk at the end [about] how [a] lot of people probably have heard of Alan Turing and the Turing Machine; where Turing in the '30s in England conceptualized modern computing and had these thought experiments in which he figured out what a computer should do, how the processing should work. And at the end of your article you talk about the fact that the actual computers that we use in real life today are really kind of deviations off that Turing idea, but the cellular computers that you've been experimenting with are actually more in keeping with Turing's original concept.
Shapiro: Yes indeed. My background is in computer science, and I have known about finite automaton Turing Machines before I learnt molecular biology. And for me when I started learning about molecular biology, it was really amazing to see the strong similarity between how living cells store information and process information and how Turing envisioned in the '30s—even before we knew about the structure of biological molecules,—how will we be doing computing device. That theoretical computing device was very much in line with how cells operate. So this inspired my search for the computing mechanism that will utilize the existing machinery that is available in the cells and enzymes that manipulate DNA can send what's written in DNA and change it; and how can we take this basic building block and reconstruct a physical machine that actually is an embodiment of the theoretical Turing Machine that Turing proposed. And what happened is they came up with a concrete design but was unable to implement it from biological molecules;
and for that they didn't have the know-how to design. So, we, with the help of engineers, actually build a mechanical plastic machine that is inspired by biological devices, but this was a start up for this research. What happened after we came up with this mechanical device, I was searching for a collaboration on and actually making it from biological molecule, and Yaakov Benenson—who is my coauthor and was my PhD student—approached me and suggested we work together. And then we started working on actual chemical implementation using his knowledge in biochemistry. And what we ended up building was not a full-fledged Turing machine, but a much more simplified device called the finite automaton, the Turing Machine just for illustration. The Turing machine can read and write information and can go left and right, and what we built was a machine that can only read information of dates and can go only in one direction and this is the finite automaton we built and reported earlier and also described in the Scientific American data.
Steve: And what actually is going on? It's hard to visualize and there is a wonderful illustrations in the article, but if you can verbally, can you try to explain what's actually chugging along in the cell?
Shapiro: Yeah! What's chugging along in the cell is essentially so simple. It's—people sometimes think or ask me, why do I call it a computer, it's just a conscious enzyme that eat up the DNA molecules; and indeed what's physically happening or chemically happening is that there is an initial DNA molecule which encodes some information in the early versions of the computer which is just very simple, you know, greater than one. In a later version we actually encoded names of disease conditions and also to the molecule at the end that we also embedded a drug molecule and what the computer is doing is actually cutting the DNA molecule one step at a time or one symbol at a time and it is cutting it in a specific way. Every time it cuts it, actually encodes piece of information—basically yes or no—and in the case of the medical application, we were describing, the question we're asking is, is there a disease? And the computer is starting with the assumption, yes there is a disease, and checks one condition at a time and keeps this bit of information on, this yes on; and as soon as it encounters the disease condition which is not present, it switches to an off state and continues processing. And at the end of this chugging along, along the DNA molecule, we know by the particular way in which the DNA molecule is cut, whether all the conditions were satisfied; and indeed there is a disease present or know one of the conditions was not satisfied and therefore we conclude that there is no disease. Once the cut completes the diagnosis part, the DNA molecule is cut in a specific way and depending on the yes or no, answer to this question. If it is yes, we'll keep cutting it till we release the drug molecule; if the answer is no, then the computation stops at this point and the drug molecule is kept at its location inactive. A lot of work remains to actually package this computer in a way that it can be delivered to a sale in computing site soon.
Steve: And it's an important point that there is an actual applied activity that you're going for. Because I think that cellular computers or molecular computers have gotten a lot of attention in the past with people thinking that we might be replacing the tower on our home computers with this box full of goop that would be performing calculations using DNA. But what you are really talking about here is using cellular machinery as a computing device to analyze the state of a living cell because those things are going to be able to naturally interact with each other.
Shapiro: What's really
is exciting for us is not a surprise to beat electronic computers—they are quite good at what they're doing—but in a way, to use computing power in the right place and in the right time to achieve the set of smart drugs. So drugs that are not just released any time, anywhere, but drugs that are released only when a disease is diagnosed at a particular location.
Steve: Great to talk to you. Thank you very much, Dr. Shapiro.
Shapiro: Thank you very much.
Steve: Shapiro's article on DNA computing is in the May issue of Scientific American and is available on our Web site, sciam.com.
We'll be right back.
Rennie: I am John Rennie, the editor in chief of Scientific American magazine. If you'd like a free preview issue of Scientific American as well as a gift, visit www.sciam.com today.
Steve: Well that's it for this edition of the Scientific American podcast. Our e-mail address is email@example.com; and also remember that science news is updated daily on the Scientific American Web site, www.sciam.com. For Science Talk, the podcast of Scientific American, I am Steve Mirsky. Thanks for clicking on us.