We just listened to a lecture on Bioinformatics by a leader in the field, Professor Pevzner. It was incredibly interesting... discussion of "genetic faults" in evolution. The lecture asked whether certain genomes are more likely to break than others, somewhat like earthquake faults. Just like California's more likely to have an earthquake than say, Illinois, are certain genomes more likely to have lasting mutations (for the species as a whole) than others?
He talked about a concept called breakpoints, which is basically what it sounds like. Points where the DNA strand breaks and rearranges itself (he discussed inversion mainly). Anyway, each rearrangement should create 2 breakpoints at most (you know... left and right...), which should be normal if evolution is random (something like... the Ohno 1973 Random SOMETHING Hypothesis). Of course, you can get one new breakpoint if one breakpoint has already been used- i.e. it's already been broken once. So if everything's random, that should be extremely unlikely. You have about 3 billion genes... what are the odds that the same places will get broken repeatedly?
However, there were 260 breakpoints found using some kind of human-mouse similarities graphing superposition and computer analysis (GRIMM maybe?), implying 130 breakpoints, assuming the random hypothesis is correct. However... some scientists managed to predict that in the 80's and confirm in the 90's that there are 245 rearrangments or something... not quite sure on the details.
So 245 is significantly greater than 130... it means that each breakpoint was used about 2 times (closer to 1.9). If you think about it... that is
extremely unlikely using the random hypothesis. There are 3 billion genes, so chance shouldn't play a big factor in it (I mean, maybe it's just coincidence if there were about 10 genes).
So that suggests that certain areas are much more likely to mutate (with a lasting effect). It's supposed to be in the July 2005 issue of
Science and written by Murphy et al, and to appear on July 22? I'm not quite sure on the details of that either...
It was incredibly interesting, everybody seemed to love it. That is a major contrast from this morning's Discovery Series Lecture (this one was arranged by our professors for our cluster, Discovery is arranged by COSMOS for all clusters). I thought the topic was actually pretty interesting. Building the Brain: From Simplicity to Complexity. The first 3/5 was actually pretty interesting. 2 sections on the basics of the brain, then the 5 steps in the development of the human embryo (which are Tissue Specification, Proliferation, Migration, Differentiation, and Formation of Connections/Synapses), then... I'm not quite sure, I couldn't stay awake. I believe something about Dynamic Development and the role of Electricity in the development of neurons. I
wanted to listen, but for some reason I just couldn't keep my eyelids open. Even if they were... the material seemed to be in way too much depth. I was nodding off into sleep all the time... I could stay awake for literally 5-10 seconds before nodding off again. But I managed to figure out that they recorded Calcium spikes (which would indicate neuron firings) and looked at the effects those had on young neurons. I managed to catch that axons would turn... The speaker was a very knowlegable guy, very enthusiastic, good speaker, interesting materials on the screen, good use of videos/iimages but I couldn't stay awake for the last half.
I guess overall the course will be interesting. The first 2.5 days weren't exactly great, but I suppose that's just some introductory stuff that we have to cover. The cluster still isn't well-organized, being the first year this has been run at COSMOS, but it should be some interesting material. I've always thought of Computer Science as programming and stuff... Mr. Hochhaus said that it was using computers to solve problems, but everyone, including himself, basically assumed that was using programming to solve programming-esque problems. Well, we've been dumped into a lot of new fields, much of it in way too much depth and which requires a background in the field to get some understanding of it. I mean... learning javascript? Requires com sci background, learning computing power in relationship to computer hardware requires hardware background, bioinfo requires bio background...
But it should be quite interesting. We're doing a lot of AI/computer perception stuff, should be fun. Gotta go now...