This is What Passes for Entertainment in Grad School - Going to Another Department's Seminar

I attended a seminar in the Math Department today. The topic was assembling genomes. Even though this is a computationally rigorous field, the speaker glazed over all the mathematically interesting details. Maybe he thought the mathematicians in the audience would be bored by the math. Because he was sponsored by the Math Department, however, he felt obligated to give a poor summary of genetics for all of the non-biologists in the audience. The mathematicians then proceeded to ask the speaker some interesting questions about biology, which was ironic considering there were at least 20 biologists in the audience (far more qualified to answer the questions). Oh, and Phil Skell was there too; he asked a question that showed absolutely no insight. More below the fold.

Mathematicians say the darndest things:

Our first question was funny because it concerns very simple math, and neither the questioner nor the speaker seemed to realize it. The questioner asked whether forensic matches between suspects and crime scenes are truly 100% identical (as they are often reported in the media). First off, this question was only tangentially related to the talk -- the speaker dealt with how assembly algorithms deal with highly similar sequences. I think this is what inspired the question, but it may have simply been the first time the mathematician had come in contact with someone who knew anything about genetics and he took the opportunity to ask the question.

So, I thought, this is a great question for a mathematician to answer; it's just simple probability. You have different versions (alleles) of rapidly evolving sequences. You sample them in a population to determine the frequencies of each of the alleles. You then genotype the suspect and the sample from the crime scene at each of the sequences. You use the frequency of the alleles to determine the probability that an individual carries a particular allele at a particular sequence. You then multiply all of the probabilities together to determine the probability of that particular genotype in the population (yes, it's a bit of an oversimplification, but it captures the gist of the procedure). That probability is the probability that a random individual would match the sample taken from the crime scene. These probabilities are usually very low (like one in a million or one in a billion) so, along with other evidence, we can be fairly certain that the match is due to the suspect being present at the crime scene and not simply due to chance. This, of course, assumes a couple of things: first, that the different loci (or sequences) are independent, and, second, that the suspect comes from the same population from which the allele frequencies are determined. Instead of describing the math, the speaker gave a sort of non-answer that was so memorable I've managed to forget it in the hour since I heard it.

The first question was off topic (with a bad answer to boot), but the second question was just straight out whack. Someone else, whom I assume was a mathematician, asked first about junk DNA. I had a nice chuckle to myself. The speaker described how, even though the vast majority of the human genome is non-protein-coding, much of it may be functional as regulatory regions or spacer sequences (he seemed obsessed with spacers for some reason). Next, she asked how humans could be most similar to chimpanzees at the DNA sequence level, but more similar to pigs at the protein level. No, that is not a typo. She really asked that question. It turns out she was confused by why we harvest pig organs for transplantation and not our closer relatives, the chimps. The speaker explained that even though we may act more like pigs (the speaker's joke, not mine), we are, in fact, closer to chimps at the protein sequence level, as well. We use pigs because of some abstract moral concepts that Janet is more qualified to explain than I am.

Phil Skell still has no idea what's going on:

I have to admit, when I saw Phil Skell walk into the room, I got very excited for the ignorant question he would ask at the end of the talk. I was disappointed. Nothing about how evolution is a historical science. Nothing about how we cannot study it experimentally. Not even a question about how similarity does not imply common ancestry. He was the first person called on at the end of the talk, and he asked about genetic difference between either humans and other species or within human populations. I don't remember the details (and I could barely here him from across the room), but it was not the kind of crazy I'd expect from the In-Skell-igent Designer.

After everyone else had asked their questions, Skell got called on a second time. His question was surprisingly on topic, but entirely non-insightful. He pointed out that even genetically identical twins may differ in the diseases they have (this is on topic because he was questioning what exactly we can learn from a whole genome sequence). Dr. Skell has realized that the genome is not the only factor that determines an organism's phenotype; gene by environment interactions determine the phenotype. Congratulations, Phil, you are now comfortable with the material from half of an introductory biology lecture. We'll cover natural selection in the next lecture.

More like this