2008 Paper 9 Question 1
Bioinformatics
(a) A long DNA sequence is used as a training set for parameter estimation of the
DNA statistical model. The observed counts of sixteen dinucleotides NXY are
as follows:


T C A G
T 306 228 126 114
C 144 102 216 138
A 222 120 132 126
G 114 102 132 132


Calculate:
(i) the transition probabilities PT T and PAG of the first-order Markov model
of the DNA sequence; [3 marks]
(ii) the transition probabilities PT T and PAG of the first-order Markov model
of the DNA sequence complementary to the given sequence. [3 marks]
(b) Build the tree from the following distance matrix between species A, B, C, D
using the UPGMA (Unweighted Pair Group Method using arithmetic
Averages) method. [7 marks]
A B C D
A 0.26 0.34 0.29
B 0.42 0.44
C 0.44
D
(c) Describe how you would build a hidden Markov model (HMM) to predict
protein secondary structure. [7 marks]