Page 26 - Applied Probability
P. 26

1. Basic Principles of Population Genetics
                                                          .
                                                          .
                                                          .
                                                                   n
                                                         =(1 − θ) [P 0 (A i B j ) − p i q j ].
                              Thus, P n (A i B j ) converges to p i q j at the geometric rate 1 − θ. For two
                              loci on different chromosomes, the deviation from linkage equilibrium is
                              halved each generation. Equilibrium is approached much more slowly for 9
                              closely spaced loci. Similar, but more cumbersome, proofs of convergence to
                              linkage equilibrium can be given for three or more loci [1, 5, 9, 11]. Problem
                              7 explores the case of three loci.
                              1.5 Selection
                              The simplest model of evolution involves selection at an autosomal locus
                              with two alleles A 1 and A 2 . At generation n, let allele A 1 have population
                              frequency p n and allele A 2 population frequency q n =1 − p n . Under the
                              usual assumptions of genetic equilibrium, we deduced the Hardy-Weinberg
                              and linkage equilibrium laws. Now suppose that we relax the assumption of
                              no selection by postulating different fitnesses w A 1 /A 1  , w A 1 /A 2  , and w A 2 /A 2
                              for the three genotypes. Fitness is a technical term dealing with the repro-
                              ductive capacity rather than the longevity of people with a given genotype.
                                                  is the ratio of the expected genetic contribution to
                              Thus, w A 1 /A 1  /w A 1 /A 2
                              the next generation of an A 1 /A 1 individual to the expected genetic con-
                              tribution of an A 1 /A 2 individual. Since only fitness ratios are relevant,
                                                                                   =1 − r, and
                              we can without loss of generality put w A 1 /A 2  =1, w A 1 /A 1
                                     =1 − s, provided of course that r ≤ 1 and s ≤ 1. Observe that r
                              w A 2 /A 2
                              and s can be negative.
                                To explore the evolutionary dynamics of this model, we define the average
                              fitness
                                                             2
                                              ¯ w n  =(1 − r)p +2p nq n +(1 − s)q 2 n
                                                             n
                                                           2
                                                  =1 − rp − sq   2 n
                                                           n
                              at generation n. Owing to our implicit assumption of random union of
                              gametes, the Hardy-Weinberg proportions appear in the definition of ¯ w n
                              even though the allele frequency p n changes over time. Because A 1 /A 1
                              individuals always contribute an A 1 allele whereas A 1 /A 2 individuals do so
                              only half of the time, the change in allele frequency ∆p n = p n+1 − p n can
                              be expressed as
                                                          2
                                                   (1 − r)p + p n q n
                                                          n
                                               =
                                          ∆p n                    − p n
                                                          ¯ w n
                                                          2               2     2
                                                   (1 − r)p + p n q n − (1 − rp − sq )p n
                                                                                n
                                                                          n
                                                          n
                                               =                                           (1.4)
                                                                   ¯ w n
                                                   p n q n [s − (r + s)p n ]
                                               =                    .
                                                           ¯ w n
   21   22   23   24   25   26   27   28   29   30   31