Page 265 - Applied Probability
P. 265

11. Radiation Hybrid Mapping
                              252
                                   Show that [9, 14]
                                                  r(1 − r)(2 − θ i )
                                           =
                                     J θ i θ i
                                              (1 − θ i r)θ i (1 − θ i + θ i r)
                                     J θ i r
                                           = J rθ i
                                                 (1 − 2r)(1 − θ i )
                                           =
                                              (1 − θ i r)(1 − θ i + θ i r)
                                                        m−1
                                                 1      	        1 − r             r
                                      J rr  =         +     θ i         +                    .
                                              r(1 − r)         r(1 − θ i r)  (1 − r)(1 − θ i + θ i r)
                                                        i=1
                                   Prove that all other entries of J are 0. Hints: Use the factorization
                                   property of the likelihood. In the case of two loci, denote the proba-
                                   bility Pr(X 1 = i, X 2 = j)by p ij for brevity. Then a typical entry J αβ
                                   of J is given by
                                                           1  1
                                                          	 	    1   ∂p ij  ∂p ij
                                                       =                         .
                                                 J αβ
                                                                 p ij  ∂α   ∂β
                                                          i=0 j=0
                                                                          has a maximum at r =  1
                                                                                              2
                                 6. Continuing the last problem, prove that J θ i θ i
                                                                                ≤ 1/[2θ i(1 − θ i )].
                                   when θ i is fixed. Use this fact to show that J θ i θ i
                                   Given a known retention probability r, this inequality proves that
                                                                                            √
                                   the asymptotic standard error of the estimated θ i will be at least  2
                                   times greater than that calculated for a simple binomial experiment
                                   with success probability θ.
                                 7. Complete the calculation of the partial derivatives of the likelihood
                                   for a single clone under the polyploid model by specifying the partial
                                   derivatives  ∂  t c,i ,  ∂  t c,i , and  ∂  α 1 (j 1 ) appearing in equations (11.16)
                                             ∂θ i   ∂r       ∂r
                                   and (11.17).
                                 8. Under the polyploid model for two loci, consider the map
                                                   (θ, r)  → (q 00 ,q 11 )
                                                          =   Pr(X 1 =0,X 2 =0)
                                                     q 00
                                                          =   Pr(X 1 =1,X 2 =1).
                                                     q 11
                                   Show that this map from {(θ, r): θ ∈ [0, 1],r ∈ (0, 1)} is one to one
                                   and onto the region
                                                                                        2
                                        Q = {(q 00 ,q 11 ): q 00 ∈ (0, 1),q 11 ∈ (0, 1),q 00 q 11 ≥ q },
                                                                                       01
                                   where q 01 =Pr(X 1 =0,X 2 = 1). Prove that θ = 0 if and only if
                                                                                   2
                                   q 00 + q 11 = 1, and θ = 1 if and only if q 00 q 11 = q . The upper
                                                                                  01
                                   boundary of Q is formed by the line q 00 + q 11 = 1 and the lower
                                                                2
                                   boundary by the curve q 00 q 11 = q . Prove that the curve is generated
                                                                01
                                                                √
                                   by the function q 11 =1 + q 00 − 2 q 00 .
   260   261   262   263   264   265   266   267   268   269   270