never use your own tissues, it's just asking for trouble.
This makes sense if you're doing tissue culture and infecting the cells with HIV in vitro or something, but I don't think there are any dangers in extracting your own DNA to use as a template for PCR. I use my own DNA for PCRs all the time. Fooling around with EtBr in dishgloves is much more dangerous.
There's a lot of stuff in there that will make you challenge how you've been thinking about science, such as the chapter about HIV and AIDS
Kary Mullis may have invented PCR, but he is still a half-insane hallucinogen user with an ax to grind. I work on HIV, and I'm all for less popular viewpoints in science, but Mullis and Duesberg's theories on AIDS pathogenesis are purely politically motivated.
Re:Not likely in the near future, but why not?
on
Linux on the Brain
·
· Score: 1
One of the problems with neural networks (both in biological and artificial systems)
One of the problems with neural networks are that people think there's anything "neural" about them at all. Backprop nets are just matrix multiplication with a nonlinear transfer function stuck in the middle (plus a method of changing the values in the matrices in response to training). This is the functional equivalent of neglecting all active membrane currents (except the ones that directly generate the action potential) and approximating the complex morphology of a neuron as a point. To be fair to researchers that use them, they have merit as a computationally tractable method for doing certain abstract nervous system modeling at the systems level (e.g. looking at receptive field sizes in visual cortex or something), but no one in neuroscience pretends that a unit in a backprop net actually behaves like a real neuron.
As far as the identical twin thing goes, we have known for a long time that the organization of the brain is not directly genetically determined. The brain is many orders of magnitude more complex than the (relatively) small size of the genome could possibly account for. Genes only tell the cells how to act in a local sense that allows the larger organization to arise (sort of like saying that the complex organization of an ant colony is not directly stored in the genome of a single ant).
Despite those small points, I very much agree that any kind of worthwhile higher-level neural interface hardware (i.e. neuromancer, not cochlear implants) would have to contain trainable networks and do a large part of the processing for you. However, I don't think that MRI is anywhere near being good enough for this (certainly not one order of magnitude away). Note that laboratory NMR scanners are already orders of magnitude better than clinical MRI scanners because you can use a 1cm sample instead of a person's head. MRI works by measuring "chemical shift" of protons, meaning how much electron density is around them. You can NOT directly measure neural activity by doing this. What you CAN measure is changes in blood flow. If a certain piece of cortex is highly active, it will (about four seconds later) swell with blood. Recently they have been able to measure the decrease in blood oxygenation that happens immediately after increased neural activity, but this is still indirect, and it still requires you to average over several trials, in both a baseline and excited state, then do complex statistics to see the trends. Even if MRI scanners weren't expensive, liquid nitrogen-cooled, the size of automobiles, and residing in heavily RF-shielded rooms, the technology would have a long ways to go.
(I've already babbled on long enough, but magnetoencephalography (MEG) is a more promising technology for that type of thing, as is sodium MRI, but these have their own problems...)
All the discussions about quantum computers seem to center around Shor's algorithm and other esoteric methods for doing things most people aren't interested in doing. Are there any algorithms for doing tasks that are actually common on desktop computers and of interest to average users?
it isn't clear how many genes one would need to change in order to enhance a trait in question. Thousands is likely much too pessimistic.
When I said "thousands" I didn't mean to imply quantitative accuracy, but also keep in mind I was not saying thousands of GENES, but thousands of polymorphic loci. That's my whole point, that phenotypes are based less on absolute presence or absence of particular genes than on their relative levels of expression (and different levels in different cell types, at different times, etc.). And each gene involved is likely to have several different combinations of promoter polymorphisms, enhancers in the introns of neighboring genes, apparently non-functional coding region polymorphisms that affect degradation or complexing of the protein with something bizarre you'd never expect, or whatever. I don't work that closely with this stuff, but someone who did could probably rattle on for pages. The problem is not unsolvable in principle, only in practice. I'm just saying that it's a LOT more complex than people outside of biology are usually aware of.
"Computational Molecular Biology" aka "Bioinformatics" is the making of algorithms used to study genetic codes
Just want to point out that computational biology and bioinformatics are not the same thing. Bioinformatics is what you're talking about with studying genetic codes, but computational biology is more along the lines of protein folding, modeling 3D structures of ribozymes, etc. It's less abstract, dealing more with predictive modeling based on actual experimentally-derived biological parameters. Don't take my word on this (I'm only a neuroscientist), but that's my understanding of it.
It always amuses me how clueless slashdot generally as group is about these things.... Despite best efforts otherwise.
I don't know about the best efforts otherwise thing. The weird thing about biology is that humans (because they're "living," I guess) have so many false notions about biology that they hold to so strongly. Biologists who don't know how to take derivatives never argue to engineers that fourier transforms are a lie (or whatever), but the opposite holds true at least 90% of the time. I used to try to explain things to people on/., but I gave up after being told again and again that I was wrong by people who don't know the difference between DNA and RNA.
As far as genetic engineering goes, genes themselves aren't nearly as important as people seem to think. Humans and chimps are supposedly 99% genetically identical, so why are the two so different? If we switched my hemoglobin with chimp hemoglobin, or slowly replaced my brain one neuron at a time with chimp neurons, I wouldn't become a chimpanzee. This is because the critical issue is not at the level of the gene products, but the REGULATION OF EXPRESSION OF THOSE GENES, especially during development. The uninformed love to spout about the wonders of genetic engineering that will unfold as soon as we have the genome sequenced, but the truth of it is that the human genome project won't solve any puzzles, it will just allow us to finally work with all the pieces.
As far as engineering genes goes, there is no intelligence gene. There is no tall gene. There is no smart gene. There are combinations of thousands of different polymorphic loci that will yield these phenotypic traits, but finding these associations is a much larger task than sequencing the genome. This is something molecular neuroscientists can't get right even today, when they make their mice that have some gene knocked out or overexpress some GluR subunit wherever, then try to draw conclusions about intelligence or aggression based on time to swim water mazes or ear bite statistics.
The real reason why human genetic engineering will not occur in our lifetimes, however, ("enhancing" humans, I mean, not fixing genetic disorders) is because humans have too many hangups about the sanctity of life. If you want to genetically engineer anything, you will inevitably have to go through hundreds and hundreds of failed attempts. Three-headed pigeon-boys with webbed legs and strange appendages. People aren't likely to volunteer their zygotes for this. Right now it's against the law to use NIH funding to obtain fetal tissue (pounds of which are THROWN AWAY at abortion clinics daily) to study AIDS, a 100% fatal disease that infects more than 11 people every minute. What are the chances that hundreds and hundreds of human lives are going to be thrown away to develop faster reflexes and 2 extra inches of penis length?
the article was (probably) trying to make the point that data in a quantum computer would not neccesaily be limited by having only two binary states available.
My understanding is that a bit in a quantum computer is still a 0 or 1, but it's both 0 and 1 simultaneously until the wave function collapses and it's forced to become one or the other. It's shocking how bad Wired is about technical details.
I was under the impression that nerves released neurotransmitters by opening their membranes. Am I mistaken?
Yes. Nerves are bundles of axons in the peripheral nervous system, and neurons (nerve cells) secrete neurotransmitters by packaging them into vesicles that fuse with the membrane, not by opening it. The great breakthroughs you are excited about (poking neurons with electrodes) occurred about 40-50 years ago.
I haven't read the real article (just the web one), but this sounds like an improved method of electroporation, a technique commonly used to perforate a cell's membrane long enough to let something slip inside (often DNA). Normally, this kills a high percentage of the cells, but this new technique sounds like it doesn't. Delivering DNA into cells is the only thing preventing us from curing diseases like cystic fibrosis or sickle-cell anemia, so this kind of thing is important.
Concerning this article I believe if we can stop apoptosis it is possible we understand the concept of aging.
Apoptosis doesn't have much to do with aging, at least not in the way you're implying. Maybe you're thinking about telomeres? (confusing bandwagons is understandable)
This "toggle switch" paper doesn't say anything about shutting off apoptosis (or any cellular gene), and even if it did, you probably wouldn't want to "shut off" apoptosis since it probably plays important roles in your body. If you wanted to test your hypothesis about apoptosis, you could easily make caspase-3 knockout mice (transgenic mice lacking a gene responsible for apoptosis) and see if they live forever (if they weren't born dead). I'm sure the mice already exist. It would be a relatively simple experiment.
What if a number of factors that inhibited/destroyed reverse transcriptase could be inserted into germline cells?
That's a really interesting idea, but I think it might be dangerous. I don't know how much is known about retrotransposons (transposable elements that use an RT to integrate the DNA copy of an RNA transcript back into genome), but some people think they have something to do with cell differentiation. Not to mention endogenous human retroviruses, about which very little is known. I think telomerase is also a form of RT, and you definitely wouldn't want to interfere with that. As far as RT inhibitors in HIV, there are non-nucleoside RT inhibitors used now in addition to AZT.
wonder what restrictions this puts on the content of the genes...
Theoretically, none. Transcriptional regulation is dependent on the DNA sequence (called the promoter) upstream of the part of the gene that codes for the protein, so you can take whatever promoter you want and attach it to whatever gene you want.
it would be necessary to fundamentally change the CD4+ T-cells' structure to cure HIV using this method
People actually think about things like this, trying to get cells to express ribozymes that cleave their own chemokine receptors, and it could be useful one day, but I don't see this as the future of HIV therapy. Viral reservoirs (where the current antivirals can't reach) are the problem with HIV, not preventing infection of T-cells (which HAART triple cocktail therapy does a reasonable job of doing already).
Nanotechnology is an area of convergence here as well
This and other reports I have read seem to suggest this, but I don't really see what this has to do with nanotechnology. If the paper discusses it, forgive me because I haven't had a chance to read it yet. But this just seems like an interesting way of controlling gene regulation. I guess the idea is you could transfect cells with something and use them as a delivery system? I see problems with this. I'm no expert in transcriptional regulation, but it's very different and a lot more complicated in eukaryotes than prokaryotes. Of course, regulating transcription of genes used in gene therapy will one day be critical (if we can ever figure out how to deliver the damn things). Luke
Who said anything about quickly? From the viewpoint of the virus, it has eternity to try and evolve a defense.
How long it takes a virus to replicate might as well be forever in terms of kinetics of the drug binding to the viral protein. What I'm saying with the bullet analogy is that there is only so much change that can occur in one generation. Over the course of 100,000 years, humans might be able to develop a hard bony plate in their chests that could stop bullets, but there is no conceivable way that a random mutation could cause something like that to arise in one generation. Someone else gave the example of throwing cats off a cliff to select for flying ones (a better example than mine). It is conceivable that a new viral protein could evolve that would somehow use a completely different mechanism and still be effective, but this is not going to happen in one generation or one hundred generations, and it will probably not happen at all if the initial resistance mutatations that arise cause the virus to be are defective. The speed issue is important because if you had a drug that killed 5% of viruses, you're leaving the other 95% to evolve new changes. If you can kill 99% of them, there are very few left to evolve. This has been shown to be important in HIV patients on HAART therapy, where if you can keep levels of viral replication low enough, significant resistance does not arise (at least over the period of years).
Viruses and bacteria evolve by almost identical mechanisms
I agree that they do at some level, but in terms of resistance to drugs, they do not, if only for the reason that antibiotics and antiviral drugs target very different things. Don't read too much into my whole viruses/bacteria distinction, my point is just that it's misguided to blindly extrapolate ideas about drug resistance from bacteria to viruses with no understanding of the underlying mechanisms. Just because bacteria developed resistance to penicillin doesn't mean that anything will magically develop resistance to any drug you throw at it. Neither bacteria or viruses have developed significant bleach resistance, for example, despite how often it is used to decontaminate surfaces.
Unfortunately, there are several reasons. First, the mechanism HIV uses for viral entry is highly complex. There is a protein on the surface of the virus called gp120 that sticks to the CD4 receptor (on a cell), which causes the shape of gp120 to change. This change in shape reveals a second site that binds to another receptor on the surface of the cell, which is actually responsible for viral entry. The two thought to be most relevant are the CXCR4 and CCR5 chemokine receptors, but HIV has been demonstrated to enter through many many other receptors as well, and even to infect cells through CD4-independent mechanisms. gp120 is a large and highly variable protein, and the binding site (which is highly variable since it sticks to many different coreceptors) we actually care about is only revealed after binding to CD4. People have been doing work along those lines, trying to get good data on the 3D structure of CD4-gp120 complexes or whatever, but it is very complicated and difficult (I haven't been following this, but I think there was a paper about it in Science this past year). The analogous data on picornaviruses was published in the early 90's (I think).
A second reason is that the body can fight off the cold, but it can't fight off HIV. From what people are saying, it doesn't sound like pleconaril is particularly effective. Shortening the cold by a couple days is one thing, but completely eradicating a virus from the body is another.
It's kind of a moot point anyway since the problem with current HIV drugs is delivery, not effectiveness. For those patients that can tolerate HAART (highly active anti-retroviral) therapy, levels of virus can be driven down below detectable levels, apparently indefinitely. However, when you take them off HAART, their bodies are full of virus in two weeks. The odd thing is that the virus that comes back isn't drug resistant virus, implying the presence of "viral reservoirs." Bob Siliciano at Johns Hopkins has done a lot of the work on the resting B-cell reservoir, and there has also been some work (by whom, I can't remember) showing that follicular dendritic cells in the lymph nodes can trap virions (free viruses) and hold them there for over 9 months in mice. People also talk about the testis/ovaries and central nervous system as reservoirs, but I'm not sure how convinced I am by that. So the challenge with HIV eradication seems not to be creating more effective anti-virals, but targeting reservoirs (not to mention figuring out ways to treat the 99% of the world who can't afford $20,000/year medications). There are some interesting ideas along these lines, but clinical trials are likely a LONG way off.
I'm not saying we will never have AI, or anything like that - I just don't believe it will be on a digital computer.
I think the human brain can be modeled in a purely digital computer, but the idea of the hardware = brain, software = consciousness analogy is all wrong. In this case, it would be hardware = laws of physics going on in the brain, software = physical structure of the brain, and certain patterns that emerge in the output = consciousness and other higher brain function. So in this sense, you're right that a digital computer could never be conscious, for example. However, a digital computer could theoretically run a program that would be. Of course, this isn't AI, this is computational neuroscience, which is my field.
As a fellow neuroscientist, I have to agree. The reason why people want to believe someone like Penrose is that people have a hard time with reducing concepts like "free will" down to a neuroscientific level. Really, I think this is because of a mistaken understanding of the relationship between "free will" and determinism. People believe that "free will" can't exist in a deterministic system. A physicist friend of mine is always trying to argue that point by saying that a murderer could go to his trial and say "Schroedinger's equation made me do it." However, determinism refers to whether or not psi_universe exists, i.e. if some omniscient ueber-being could predict everything that will ever happen. "Free will" refers to decisions that humans make. The making of a decision is something that happens when the physical state of your brain goes one way and not another -- your brain is PART of this whole system. So determinism refers to whether or not the outcome of your decision would have been theoretically predictable, which is not the same as whether or not "free will" operated in the making of that decision.
As soon as you introduce a drug which interferes with a function, you have placed selective pressure on that protein to evolve a new mechanism.
That's true, but evolution isn't some magic thing that can defeat all obstacles. Saying that a viral protein could quickly evolve into one that works in an entirely different way is analogous to saying that filling rooms full of people and shooting them will select for bullet-resistant humans.
That's how we have Vancomycin resistant strains now.
I think that by the time you are sick enough that you need the drug, you will already have antibodies to it, so I don't see that as being a serious issue.
Everyone seems to be worried that using pleconaril will generate some kind of super-viruses that are impossible to kill, and I think this stems from confusion about a few things. The primary thing is that viruses are not bacteria. Bacteria are complex organisms that gain resistance through a variety of mechanisms. Overuse of antibiotics leading to resistant strains of bacteria is a very different issue. Viruses are much simpler than bacteria, and resistance often arises through mutant forms of viral proteins (they don't have large genomes to hold new genes that confer resistance). The reason why resistance mutations can be less of a problem with viruses is because these mutant proteins don't always function properly. Drugs like pleconaril are designed to target parts of the protein that, if mutated, would render it semi-functional, leading to a weakened virus. Attenuated virulence has been reported in some drug-resistant HIV strains, and (as I just checked in medline) in pleconaril-resistant viruses as well. I'm not saying drug resistance isn't ever a problem, just that it's likely that pleconaril will not become ineffective after two years on the market.
The promise of carefully designed drugs is that we can keep pace with evolution better. Much drug development is still done with a shotgun.
That's part of the significance of this type of drug, that it was specifically designed based on knowledge of the 3D structure of a viral protein. Resistance mutations will obviously occur, but these mutations can only go so far. If the drug is intelligently designed and binds to a part of the protein that is critical for its proper function, for example, resistance mutations will give rise to defective viruses. I read a while ago that this had been observed by people developing neuraminidase inhibitors; I'm not sure if this new drug might be one of those (I'll have to read the paper).
In terms of resistance mutations, it's also important not to confuse viruses with bacteria, which are much more complex and become resistant in different ways.
The patenting of genes or such a process will make the information available publicly
Do you propose instead that all biological scientists keep their data secret? Progress would grind to a halt.
will they really be controlled by this absurd notion of the Patent?
I think that's the real question, if patents are the best method to use. The processes involved are general-purpose techniques that can be used for anything from treating cancer to creating deadlier viruses, and even if they're patentable (like PCR), they're simple enough that a smart undergrad could implement them, so the real issue is money, not secrecy. As far as the genes themselves, data needs to be shared if any progress is to be made, and patents allow the people who made the discoveries to benefit from their application.
most of us don't really understand how the Genetic field works
I don't consider myself a molecular biologist by any stretch of the imagination, just someone who uses molecular biology for neuroscience, but I know that it is true that a lot of the opposition to things like gene patenting comes from a misunderstanding of how research in these fields actually works.
First, it's not only companies that patent genes, it's individuals in universities and other research institutions. More importantly, discovery of genes is generally not as directed as one would think. In a single week, there are probably hundreds of novel genes discovered in labs across the world, mostly by people who don't necessarily have any idea what the gene products do (there are different techniques used, but the ones I'm familiar with do it either on the basis of interactions between the novel gene product and a known protein or homology to known genes).
In other words, the people who find the genes are not the same people who figure out what to do with them. Monsanto might want to make an Alzheimer's drug that inhibits an enzyme whose gene was first sequenced by a postdoc in the nephrology department at the Medical College of Wisconsin.
Gene patents are good because they reward a gene's discoverer if a big company comes along with their million dollar screening facilities and makes billions off that person's insight. It also facilitates the sharing of data, which might otherwise be kept secret and sold to the highest bidder.
I don't claim that patents are necessarily the best way to do this, but something has to fill this role.
Slashdot Mirror
never use your own tissues, it's just asking for trouble.
This makes sense if you're doing tissue culture and infecting the cells with HIV in vitro or something, but I don't think there are any dangers in extracting your own DNA to use as a template for PCR. I use my own DNA for PCRs all the time. Fooling around with EtBr in dishgloves is much more dangerous.
There's a lot of stuff in there that will make you challenge how you've been thinking about science, such as the chapter about HIV and AIDS
Kary Mullis may have invented PCR, but he is still a half-insane hallucinogen user with an ax to grind. I work on HIV, and I'm all for less popular viewpoints in science, but Mullis and Duesberg's theories on AIDS pathogenesis are purely politically motivated.
One of the problems with neural networks (both in biological and artificial systems)
One of the problems with neural networks are that people think there's anything "neural" about them at all. Backprop nets are just matrix multiplication with a nonlinear transfer function stuck in the middle (plus a method of changing the values in the matrices in response to training). This is the functional equivalent of neglecting all active membrane currents (except the ones that directly generate the action potential) and approximating the complex morphology of a neuron as a point. To be fair to researchers that use them, they have merit as a computationally tractable method for doing certain abstract nervous system modeling at the systems level (e.g. looking at receptive field sizes in visual cortex or something), but no one in neuroscience pretends that a unit in a backprop net actually behaves like a real neuron.
As far as the identical twin thing goes, we have known for a long time that the organization of the brain is not directly genetically determined. The brain is many orders of magnitude more complex than the (relatively) small size of the genome could possibly account for. Genes only tell the cells how to act in a local sense that allows the larger organization to arise (sort of like saying that the complex organization of an ant colony is not directly stored in the genome of a single ant).
Despite those small points, I very much agree that any kind of worthwhile higher-level neural interface hardware (i.e. neuromancer, not cochlear implants) would have to contain trainable networks and do a large part of the processing for you. However, I don't think that MRI is anywhere near being good enough for this (certainly not one order of magnitude away). Note that laboratory NMR scanners are already orders of magnitude better than clinical MRI scanners because you can use a 1cm sample instead of a person's head. MRI works by measuring "chemical shift" of protons, meaning how much electron density is around them. You can NOT directly measure neural activity by doing this. What you CAN measure is changes in blood flow. If a certain piece of cortex is highly active, it will (about four seconds later) swell with blood. Recently they have been able to measure the decrease in blood oxygenation that happens immediately after increased neural activity, but this is still indirect, and it still requires you to average over several trials, in both a baseline and excited state, then do complex statistics to see the trends. Even if MRI scanners weren't expensive, liquid nitrogen-cooled, the size of automobiles, and residing in heavily RF-shielded rooms, the technology would have a long ways to go.
(I've already babbled on long enough, but magnetoencephalography (MEG) is a more promising technology for that type of thing, as is sodium MRI, but these have their own problems...)
All the discussions about quantum computers seem to center around Shor's algorithm and other esoteric methods for doing things most people aren't interested in doing. Are there any algorithms for doing tasks that are actually common on desktop computers and of interest to average users?
it isn't clear how many genes one would need to change in order to enhance a trait in question. Thousands is likely much too pessimistic.
When I said "thousands" I didn't mean to imply quantitative accuracy, but also keep in mind I was not saying thousands of GENES, but thousands of polymorphic loci. That's my whole point, that phenotypes are based less on absolute presence or absence of particular genes than on their relative levels of expression (and different levels in different cell types, at different times, etc.). And each gene involved is likely to have several different combinations of promoter polymorphisms, enhancers in the introns of neighboring genes, apparently non-functional coding region polymorphisms that affect degradation or complexing of the protein with something bizarre you'd never expect, or whatever. I don't work that closely with this stuff, but someone who did could probably rattle on for pages. The problem is not unsolvable in principle, only in practice. I'm just saying that it's a LOT more complex than people outside of biology are usually aware of.
"Computational Molecular Biology" aka "Bioinformatics" is the making of algorithms used to study genetic codes
Just want to point out that computational biology and bioinformatics are not the same thing. Bioinformatics is what you're talking about with studying genetic codes, but computational biology is more along the lines of protein folding, modeling 3D structures of ribozymes, etc. It's less abstract, dealing more with predictive modeling based on actual experimentally-derived biological parameters. Don't take my word on this (I'm only a neuroscientist), but that's my understanding of it.
It always amuses me how clueless slashdot generally as group is about these things.... Despite best efforts otherwise.
/., but I gave up after being told again and again that I was wrong by people who don't know the difference between DNA and RNA.
I don't know about the best efforts otherwise thing. The weird thing about biology is that humans (because they're "living," I guess) have so many false notions about biology that they hold to so strongly. Biologists who don't know how to take derivatives never argue to engineers that fourier transforms are a lie (or whatever), but the opposite holds true at least 90% of the time. I used to try to explain things to people on
As far as genetic engineering goes, genes themselves aren't nearly as important as people seem to think. Humans and chimps are supposedly 99% genetically identical, so why are the two so different? If we switched my hemoglobin with chimp hemoglobin, or slowly replaced my brain one neuron at a time with chimp neurons, I wouldn't become a chimpanzee. This is because the critical issue is not at the level of the gene products, but the REGULATION OF EXPRESSION OF THOSE GENES, especially during development. The uninformed love to spout about the wonders of genetic engineering that will unfold as soon as we have the genome sequenced, but the truth of it is that the human genome project won't solve any puzzles, it will just allow us to finally work with all the pieces.
As far as engineering genes goes, there is no intelligence gene. There is no tall gene. There is no smart gene. There are combinations of thousands of different polymorphic loci that will yield these phenotypic traits, but finding these associations is a much larger task than sequencing the genome. This is something molecular neuroscientists can't get right even today, when they make their mice that have some gene knocked out or overexpress some GluR subunit wherever, then try to draw conclusions about intelligence or aggression based on time to swim water mazes or ear bite statistics.
The real reason why human genetic engineering will not occur in our lifetimes, however, ("enhancing" humans, I mean, not fixing genetic disorders) is because humans have too many hangups about the sanctity of life. If you want to genetically engineer anything, you will inevitably have to go through hundreds and hundreds of failed attempts. Three-headed pigeon-boys with webbed legs and strange appendages. People aren't likely to volunteer their zygotes for this. Right now it's against the law to use NIH funding to obtain fetal tissue (pounds of which are THROWN AWAY at abortion clinics daily) to study AIDS, a 100% fatal disease that infects more than 11 people every minute. What are the chances that hundreds and hundreds of human lives are going to be thrown away to develop faster reflexes and 2 extra inches of penis length?
the article was (probably) trying to make the point that data in a quantum computer would not neccesaily be limited by having only two binary states available.
My understanding is that a bit in a quantum computer is still a 0 or 1, but it's both 0 and 1 simultaneously until the wave function collapses and it's forced to become one or the other. It's shocking how bad Wired is about technical details.
I was under the impression that nerves released neurotransmitters by opening their membranes. Am I mistaken?
Yes. Nerves are bundles of axons in the peripheral nervous system, and neurons (nerve cells) secrete neurotransmitters by packaging them into vesicles that fuse with the membrane, not by opening it. The great breakthroughs you are excited about (poking neurons with electrodes) occurred about 40-50 years ago.
I haven't read the real article (just the web one), but this sounds like an improved method of electroporation, a technique commonly used to perforate a cell's membrane long enough to let something slip inside (often DNA). Normally, this kills a high percentage of the cells, but this new technique sounds like it doesn't. Delivering DNA into cells is the only thing preventing us from curing diseases like cystic fibrosis or sickle-cell anemia, so this kind of thing is important.
Concerning this article I believe if we can stop apoptosis it is possible we understand the concept of aging.
Apoptosis doesn't have much to do with aging, at least not in the way you're implying. Maybe you're thinking about telomeres? (confusing bandwagons is understandable)
This "toggle switch" paper doesn't say anything about shutting off apoptosis (or any cellular gene), and even if it did, you probably wouldn't want to "shut off" apoptosis since it probably plays important roles in your body. If you wanted to test your hypothesis about apoptosis, you could easily make caspase-3 knockout mice (transgenic mice lacking a gene responsible for apoptosis) and see if they live forever (if they weren't born dead). I'm sure the mice already exist. It would be a relatively simple experiment.
Luke
What if a number of factors that inhibited/destroyed reverse transcriptase could be inserted into germline cells?
That's a really interesting idea, but I think it might be dangerous. I don't know how much is known about retrotransposons (transposable elements that use an RT to integrate the DNA copy of an RNA transcript back into genome), but some people think they have something to do with cell differentiation. Not to mention endogenous human retroviruses, about which very little is known. I think telomerase is also a form of RT, and you definitely wouldn't want to interfere with that. As far as RT inhibitors in HIV, there are non-nucleoside RT inhibitors used now in addition to AZT.
Luke
wonder what restrictions this puts on the content of the genes...
Theoretically, none. Transcriptional regulation is dependent on the DNA sequence (called the promoter) upstream of the part of the gene that codes for the protein, so you can take whatever promoter you want and attach it to whatever gene you want.
it would be necessary to fundamentally change the CD4+ T-cells' structure to cure HIV using this method
People actually think about things like this, trying to get cells to express ribozymes that cleave their own chemokine receptors, and it could be useful one day, but I don't see this as the future of HIV therapy. Viral reservoirs (where the current antivirals can't reach) are the problem with HIV, not preventing infection of T-cells (which HAART triple cocktail therapy does a reasonable job of doing already).
Nanotechnology is an area of convergence here as well
This and other reports I have read seem to suggest this, but I don't really see what this has to do with nanotechnology. If the paper discusses it, forgive me because I haven't had a chance to read it yet. But this just seems like an interesting way of controlling gene regulation. I guess the idea is you could transfect cells with something and use them as a delivery system? I see problems with this. I'm no expert in transcriptional regulation, but it's very different and a lot more complicated in eukaryotes than prokaryotes. Of course, regulating transcription of genes used in gene therapy will one day be critical (if we can ever figure out how to deliver the damn things). Luke
blindly extrapolate ideas about drug resistance from bacteria to viruses with no understanding of the underlying mechanisms
And I didn't mean to accuse you of that, just almost everyone else here.
Who said anything about quickly? From the viewpoint of the virus, it has eternity to try and evolve a defense.
How long it takes a virus to replicate might as well be forever in terms of kinetics of the drug binding to the viral protein. What I'm saying with the bullet analogy is that there is only so much change that can occur in one generation. Over the course of 100,000 years, humans might be able to develop a hard bony plate in their chests that could stop bullets, but there is no conceivable way that a random mutation could cause something like that to arise in one generation. Someone else gave the example of throwing cats off a cliff to select for flying ones (a better example than mine). It is conceivable that a new viral protein could evolve that would somehow use a completely different mechanism and still be effective, but this is not going to happen in one generation or one hundred generations, and it will probably not happen at all if the initial resistance mutatations that arise cause the virus to be are defective. The speed issue is important because if you had a drug that killed 5% of viruses, you're leaving the other 95% to evolve new changes. If you can kill 99% of them, there are very few left to evolve. This has been shown to be important in HIV patients on HAART therapy, where if you can keep levels of viral replication low enough, significant resistance does not arise (at least over the period of years).
Viruses and bacteria evolve by almost identical mechanisms
I agree that they do at some level, but in terms of resistance to drugs, they do not, if only for the reason that antibiotics and antiviral drugs target very different things. Don't read too much into my whole viruses/bacteria distinction, my point is just that it's misguided to blindly extrapolate ideas about drug resistance from bacteria to viruses with no understanding of the underlying mechanisms. Just because bacteria developed resistance to penicillin doesn't mean that anything will magically develop resistance to any drug you throw at it. Neither bacteria or viruses have developed significant bleach resistance, for example, despite how often it is used to decontaminate surfaces.
Unfortunately, there are several reasons. First, the mechanism HIV uses for viral entry is highly complex. There is a protein on the surface of the virus called gp120 that sticks to the CD4 receptor (on a cell), which causes the shape of gp120 to change. This change in shape reveals a second site that binds to another receptor on the surface of the cell, which is actually responsible for viral entry. The two thought to be most relevant are the CXCR4 and CCR5 chemokine receptors, but HIV has been demonstrated to enter through many many other receptors as well, and even to infect cells through CD4-independent mechanisms. gp120 is a large and highly variable protein, and the binding site (which is highly variable since it sticks to many different coreceptors) we actually care about is only revealed after binding to CD4. People have been doing work along those lines, trying to get good data on the 3D structure of CD4-gp120 complexes or whatever, but it is very complicated and difficult (I haven't been following this, but I think there was a paper about it in Science this past year). The analogous data on picornaviruses was published in the early 90's (I think).
A second reason is that the body can fight off the cold, but it can't fight off HIV. From what people are saying, it doesn't sound like pleconaril is particularly effective. Shortening the cold by a couple days is one thing, but completely eradicating a virus from the body is another.
It's kind of a moot point anyway since the problem with current HIV drugs is delivery, not effectiveness. For those patients that can tolerate HAART (highly active anti-retroviral) therapy, levels of virus can be driven down below detectable levels, apparently indefinitely. However, when you take them off HAART, their bodies are full of virus in two weeks. The odd thing is that the virus that comes back isn't drug resistant virus, implying the presence of "viral reservoirs." Bob Siliciano at Johns Hopkins has done a lot of the work on the resting B-cell reservoir, and there has also been some work (by whom, I can't remember) showing that follicular dendritic cells in the lymph nodes can trap virions (free viruses) and hold them there for over 9 months in mice. People also talk about the testis/ovaries and central nervous system as reservoirs, but I'm not sure how convinced I am by that. So the challenge with HIV eradication seems not to be creating more effective anti-virals, but targeting reservoirs (not to mention figuring out ways to treat the 99% of the world who can't afford $20,000/year medications). There are some interesting ideas along these lines, but clinical trials are likely a LONG way off.
Luke
I'm not saying we will never have AI, or anything like that - I just don't believe it will be on a digital computer.
I think the human brain can be modeled in a purely digital computer, but the idea of the hardware = brain, software = consciousness analogy is all wrong. In this case, it would be hardware = laws of physics going on in the brain, software = physical structure of the brain, and certain patterns that emerge in the output = consciousness and other higher brain function. So in this sense, you're right that a digital computer could never be conscious, for example. However, a digital computer could theoretically run a program that would be. Of course, this isn't AI, this is computational neuroscience, which is my field.
As a fellow neuroscientist, I have to agree. The reason why people want to believe someone like Penrose is that people have a hard time with reducing concepts like "free will" down to a neuroscientific level. Really, I think this is because of a mistaken understanding of the relationship between "free will" and determinism. People believe that "free will" can't exist in a deterministic system. A physicist friend of mine is always trying to argue that point by saying that a murderer could go to his trial and say "Schroedinger's equation made me do it." However, determinism refers to whether or not psi_universe exists, i.e. if some omniscient ueber-being could predict everything that will ever happen. "Free will" refers to decisions that humans make. The making of a decision is something that happens when the physical state of your brain goes one way and not another -- your brain is PART of this whole system. So determinism refers to whether or not the outcome of your decision would have been theoretically predictable, which is not the same as whether or not "free will" operated in the making of that decision.
As soon as you introduce a drug which interferes with a function, you have placed selective pressure on that protein to evolve a new mechanism.
That's true, but evolution isn't some magic thing that can defeat all obstacles. Saying that a viral protein could quickly evolve into one that works in an entirely different way is analogous to saying that filling rooms full of people and shooting them will select for bullet-resistant humans.
That's how we have Vancomycin resistant strains now.
That's for bacteria, not viruses.
I think that by the time you are sick enough that you need the drug, you will already have antibodies to it, so I don't see that as being a serious issue.
Everyone seems to be worried that using pleconaril will generate some kind of super-viruses that are impossible to kill, and I think this stems from confusion about a few things. The primary thing is that viruses are not bacteria. Bacteria are complex organisms that gain resistance through a variety of mechanisms. Overuse of antibiotics leading to resistant strains of bacteria is a very different issue. Viruses are much simpler than bacteria, and resistance often arises through mutant forms of viral proteins (they don't have large genomes to hold new genes that confer resistance). The reason why resistance mutations can be less of a problem with viruses is because these mutant proteins don't always function properly. Drugs like pleconaril are designed to target parts of the protein that, if mutated, would render it semi-functional, leading to a weakened virus. Attenuated virulence has been reported in some drug-resistant HIV strains, and (as I just checked in medline) in pleconaril-resistant viruses as well. I'm not saying drug resistance isn't ever a problem, just that it's likely that pleconaril will not become ineffective after two years on the market.
The promise of carefully designed drugs is that we can keep pace with evolution better. Much drug development is still done with a shotgun.
That's part of the significance of this type of drug, that it was specifically designed based on knowledge of the 3D structure of a viral protein. Resistance mutations will obviously occur, but these mutations can only go so far. If the drug is intelligently designed and binds to a part of the protein that is critical for its proper function, for example, resistance mutations will give rise to defective viruses. I read a while ago that this had been observed by people developing neuraminidase inhibitors; I'm not sure if this new drug might be one of those (I'll have to read the paper).
In terms of resistance mutations, it's also important not to confuse viruses with bacteria, which are much more complex and become resistant in different ways.
The patenting of genes or such a process will make the information available publicly
Do you propose instead that all biological scientists keep their data secret? Progress would grind to a halt.
will they really be controlled by this absurd notion of the Patent?
I think that's the real question, if patents are the best method to use. The processes involved are general-purpose techniques that can be used for anything from treating cancer to creating deadlier viruses, and even if they're patentable (like PCR), they're simple enough that a smart undergrad could implement them, so the real issue is money, not secrecy. As far as the genes themselves, data needs to be shared if any progress is to be made, and patents allow the people who made the discoveries to benefit from their application.
most of us don't really understand how the Genetic field works
I don't consider myself a molecular biologist by any stretch of the imagination, just someone who uses molecular biology for neuroscience, but I know that it is true that a lot of the opposition to things like gene patenting comes from a misunderstanding of how research in these fields actually works.
First, it's not only companies that patent genes, it's individuals in universities and other research institutions. More importantly, discovery of genes is generally not as directed as one would think. In a single week, there are probably hundreds of novel genes discovered in labs across the world, mostly by people who don't necessarily have any idea what the gene products do (there are different techniques used, but the ones I'm familiar with do it either on the basis of interactions between the novel gene product and a known protein or homology to known genes).
In other words, the people who find the genes are not the same people who figure out what to do with them. Monsanto might want to make an Alzheimer's drug that inhibits an enzyme whose gene was first sequenced by a postdoc in the nephrology department at the Medical College of Wisconsin.
Gene patents are good because they reward a gene's discoverer if a big company comes along with their million dollar screening facilities and makes billions off that person's insight. It also facilitates the sharing of data, which might otherwise be kept secret and sold to the highest bidder.
I don't claim that patents are necessarily the best way to do this, but something has to fill this role.