Math Grad Seminar Oct 11, 2006

How Can One Tell In What Direction Evolution is Going?

Stanley Sawyer

Abstract:
In the long run, most important changes in biological organisms are due to the replacements of genes by new genes that do a better job for the organism. However, in large, established populations, many biologists believe that most evolutionary change is, instead, due to the replacement of genes by slightly deleterious variants.

The reason for this is that most mutations are harmful rather than helpful, and that mildly harmful mutations can replace a better, established gene by the chance effects of who mates with whom and who happens to survive. This would take a long time for a large population, but most large populations have been around for a very long time. We can study this question from the distribution of DNA in populations in the present. The distribution of a new gene is different if it is advantageous, deleterious, or selectively the same as the old gene. We can obtain more information from the differences between two related species. Current data suggests that most recent evolutionary changes in fruit flies were advantageous, but that most recent evolutionary changes in a common weed were deleterious. Investigating this process in detail leads to difficult problems in probability theory and statistics, as well as to the use of a technique called Markov Chain Monte Carlo.

Transparencies for Talk   (PDF format)
(Some genetics, some probability, some statistics, using MCMC to estimate parameters)

The main parameters:

Mugamma: The overall mean of selection coefficients over 73 genetic loci
Wsigma: The standard deviation of selection coefficients of new mutants within loci
Bsigma: The standard deviation of locus means between loci

Graphics for the Talk:

The following are plots using 5000 values from runs of a high-dimensional Markov Chain
    with 21,000,000 or 5,500,000 iterations:

The three main parameters have a TRIMODAL distribution in the full run:

Histogram of Wsigma
Scatterplot of Mugamma x Wsigma
Scatterplot of Mugamma x Bsigma
The full run goes back and forth between two multimodes:
Trace Plot of Mugamma
Using an informative prior to block out the third mode:
Now there are only two mildly overlapping modes:
Histogram of Wsigma
Trace Plot of Wgamma


Return to Stanley Sawyer non-course handout home page

  • Click here to return to Stanley Sawyer's home page
  • Click here for the Mathematics Department home page

    Last modified November 1, 2006