Information Technology Reference
In-Depth Information
Table 14.4
(continued)
Sequence
Similarity
EX.6
A1
A2
A3
A4
A1
A2
A3
A4
0.0
A1
0
3.85
3.21
2.16
0
1
3
2
1.0
A2
3.84
0
3.01
3.37
3
0
1
2
0.0
A3
3.38
2.85
0
3.46
3
1
0
2
0.0
A4
2.11
3.17
3.44
0
2
3
1
0
Sequence
Similarity
EX.7
A1
A2
A3
A4
A1
A2
A3
A4
0.0
A1
0
2.17
3.41
3.30
0
2
3
1
0.0
A2
1.84
0
2.11
3.97
2
0
1
3
1.0
3.44
0
3.10
3
0
2
A3
2.02
1
0.0
A4
3.33
3.97
3.63
0
2
3
1
0
9.5/24
= 0.40
Using the cumulative binomial probability calculations (Feller
1968
) to assess
the confidence of these results, we find that the probability of getting eight or more
correct sequences is 0.035444 or less than 4 % of the time. For nine or more correct
sequences it drops to 0.01176 or slightly greater than 1 % and ten or more it is
0.003339 or a slightly higher than 0.3 %. Since we have predicted 9.5 sequences
correctly (using 0.5 where not all information is available) we can be confident in
our model to about 1 % that the predictions can happen by chance. This is quite
good.
14.8
Ally Choice
The results seem to indicate that the frequency of ally choice order relates to the
relative z-score distances calculated. The complete sequences were not always given
but where a particular position is given we have included this in Figs.
14.6
and
14.7
.
We can compare the shape of the cumulative graphs for relative z-score distance
for each position (first, second and third) for both the observed and the model. The
model is on the left and the observed results on the right:
