Page 149 - Applied Probability
P. 149

133
                                                      7. Computation of Mendelian Likelihoods
                                Inspection of the location score curve shows that it rises above the mag-
                              ical level of 3 and that the episodic ataxia gene probably resides on the
                              interval from D12S372 to pY2/1. Where the marker S372 is uninformative
                              for linkage, other markers fill the information gap. Thus, location scores
                              make better use of scarce disease pedigrees than lod scores do. In Chapter
                              9 we will revisit this problem and demonstrate how the haplotypes dis-
                              played in Figure 7.3 are reconstructed and how one can compute location
                              scores using an almost arbitrary number of markers.
                              7.7 Problems
                                 1. Under Haldane’s model of independent recombination on disjoint in-
                                   tervals, it is possible to compute the recombination fraction θ ij be-
                                   tween two loci i< j by Trow’s formula

                                                                j−1

                                                    1 − 2θ ij  =   (1 − 2θ k,k+1 ),       (7.10)
                                                                k=i
                                   where the loci occur in numerical order along the chromosome, and
                                   where θ k,k+1 is the recombination fraction between the adjacent loci
                                   k and k + 1. Verify Trow’s formula first for three loci (i = 1 and
                                   j = 3) and then by induction for an arbitrary number of loci. (Hint:
                                   For three loci, recombination occurs between loci 1 and 3 if and only
                                   if it occurs between loci 1 and 2 and not between loci 2 and 3, or vice
                                   versa.)

                                 2. Consider the partially typed, inbred pedigree depicted in Figure 7.6.
                                   The phenotypes displayed in the figure are unordered genotypes at a
                                   single codominant locus with three alleles. Show that the genotype
                                   elimination algorithm fails to eliminate some superfluous genotypes
                                   in this pedigree.

                                 3. The sum of array products


                                                    ···     A(G 1 ,G 2 ,G 3 ,G 4 )B(G 4 ,G 5 )
                                               G 1 ∈S 1  G 9 ∈S 9
                                                             × C(G 5 ,G 6 )D(G 6 ,G 7 ,G 8 ,G 9 )

                                   can be evaluated as an iterated sum by the greedy algorithm. If all
                                   range sets S i have the same number of elements m> 2, then show
                                   that one greedy summation sequence is (5, 1, 2, 3, 4, 7, 8, 9, 6). Prove
                                   that the alternative nongreedy sequence (1, 2, 3, 4, 5, 7, 8, 9, 6) requires
                                   fewer arithmetic operations (additions plus multiplications) [18].
   144   145   146   147   148   149   150   151   152   153   154