Slashdot Mirror
Modeling the Building Blocks of Life
eldavojohn writes "A new research paper is creating some buzz about the roles of computer engineering in biology. Historically, computational techniques in genome sequencing have proved useful in predicting which DNA sequence produces which amino acid and which amino acid sequence produces which protein. Now, this new research is leading towards a robust model of proteins and their messaging systems. This is one step further in understanding the basics of life and, consequently, pushes us closer to being able to emulate organisms entirely from the bottom up instead of our failed prior approaches of from the top down. A long way from perfect, but an opening into a wide field of study and maybe even a new division of biology."
This may be a bit picky, but the work being done here is not computer engineering but rather computer science. Computer engineering generally refers to engineering techniques for building computers and computer systems (including parts of electrical engineering, materials science, algorithms, etc.) whereas computer science is the study of algorithms. This work is not designing computing systems but rather using algorithms to model the building blocks of biology.
http://yetanotherpoliticalrant.blogspot.com
Usually when computers and numerical modeling techniques are used to understand and solve problems in sciences the term we use is computational. Using computers in fulid mechanics, it is computational fluid mechanics. Similarly there are computational electromagnetics, computational solid mechanics (usually finite element methods) computational geometry etc. So in that way, the correct term here would have been, "computational biology".
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
http://en.wikipedia.org/wiki/Grafting
How about bioinformatics?
I used to carry a bottle of whiskey for snake bite. And two snakes. -Nefarious Wheel
you think so? How many animals are really studied do you think? I can give you the answer: C. elegans, fruitfly, zebrafish, bulfrog, chicken, mice and rats. The rest hardly counts. The reason why they choose C. elegans in this case you already stated, the principles of signal transduction apply just as much to human beings as to those worms. Your EGFR/RAS/MAPK pathway isn't that different...
While this project might be interesting to some, this is hardly a new approach to biology.
/. summary, but it does not really give the reader a good idea of the current state of the HUGE field of computational biology!
Computational Biology has been around for quite a while now and simulation is actually one of its strongest points so far.
There used to only be two main settings for conducting experiments: in vitro (outside of living organisms, literally within a glass) and in vivo (done in living tissue/organisms).
With the advent of comp. bio., a new and comparatively incredibly inexpensive way of experimenting has become available: in silico (experiments are simulated) This is pretty much what the article was talking about and has been a massive success in biology, for quite some time now!
Since this term has been used since the 1990s, this is not exactly new.
I won't even go into talking about the misleading
Interesting idea but I would reserve a few questions for such a proposition. First of all, how would you create a virus that doesn't mutate? Viruses that go through the lysogenic cycle(AIDS for example), by definition incorporate into host DNA to hijack the processes of cellular protein production. This is an inherently mutation prone process especially during separation from host DNA. I suppose adding an error checking protein functionally comparable to some polymerases for dna replication would be possible, but at present we are somewhat far off from being able to design a protein to perform an arbitrary function that doesn't already exist in another organism. Supposing you could do all that somehow, how do you propose to computationally predict a "combination that works" against the next type of mutation when mutation itself is a completely random and unpredictable process?
A mutation-resistant virus is easy enough. dsDNA viruses are quite stable, and most of them function by adding their own chromosomes to the nucleus, without altering the host DNA. The result is that they rely on the same DNA polymerase and proofreading enzymes that the cell uses for its own replication. The poxviruses, for instance, are a reasonable template for fashioning an HIV counter-virus, as they generally replicate in this manner.
One major challenge would be figuring out how to reliably recognize an HIV infection-in-progress: since HIV is a ssRNA retrovirus, it mutates quite rapidly. There are very few genes or proteins that can be reliably targeted, but portions of the HIV genome are conserved, so it's not hopeless. (HIV mutates so rapidly that the grandparent poster's suggestion of adapting to new mutations using computer simulations is unrealistic. In a single HIV-infected patient, new minor strains emerge on a daily basis.)
That raises the question of what to do once an HIV infection is encountered. Since HIV (as a retrovirus) operates by irreversibly incorporating its genome into the host DNA, any counter-virus would presumably operate by triggering apoptosis when it co-infected a cell with HIV. An alternative would be to silence HIV gene expression, rendering the cell infected but dormant and non-infectious.
Another major challenge would be replication. Most viruses destroy the host cell when they replicate. Even the ones that don't rip the cell membrane open will generally deplete the cell's energy reserves and kill it indirectly. The counter-virus replication rate would be a careful balance between ensuring that there are enough viral particles to infect all HIV-infected cells, versus taxing the HIV-free host cells to the point of causing disease.
Another challenge would be immune system evasion. Since one goal is for the counter-virus to have a low mutation rate (to prevent it from becoming a disease in its own right) the counter-virus would be ill-equipped to evade the adaptive immune system. One option would be to code for the viral coat on a highly variable ssRNA strand, while the replication and counter-viral genes remain on more stable dsDNA; such mixed-genome viruses, although rare, do exist in nature. Another option would be to investigate Adeno-Associated Virus, which the immune system essentially gives a free pass to since it is so "well-behaved"; the counter-virus could mimic AAV's viral coat and recognition signals. The adaptive immune system eventually targets even AAV, but since HIV attacks the adaptive immune system, the mimicry might be enough to buy time for the counter-virus to put a serious dent in the HIV count.
It'd be difficult in the extreme, and it would require years of computer time to figure out how everything fits together, but it's at least conceivable with current technology to modify an existing virus for the purpose.
Range Voting: preference intensity matters