|
Relative Count
|
Natural Parameter
|
Maximum a posteriori
|
---|
- | A | C | G | T | A | C | G | T | A | C | G | T |
---|
A
| 97.46 | 0.10 | 0.24 | 1.15 | 97.46 | 0.10 | 0.24 | 1.15 | 97.47 | 0.13 | 0.20 | 1.14 |
C
| 0.27 | 99.72 | 0.01 | 1.16 | 0.27 | 99.72 | 0.01 | 1.16 | 0.24 | 99.66 | 0.02 | 1.17 |
G
| 1.16 | 0.02 | 99.58 | 0.18 | 1.16 | 0.02 | 99.58 | 0.18 | 1.16 | 0.02 | 99.65 | 0.23 |
T
| 1.11 | 0.16 | 0.17 | 97.51 | 1.11 | 0.16 | 0.17 | 97.51 | 1.12 | 0.19 | 0.13 | 97.46 |
- Three estimates for the expected frequency of nucleotide mutation P as a function of the counts C in Table 1 given as relative counts with p
ij
= C
ij
/ Σ
i
C
ij
, natural parameter means as per equation (13), and maximum a posteriori estimates when constrained by the polymerase misincorporation model with frequencies T. The relative-count and natural-parameter frequencies are virtually identical, while the polymerase model-constrained estimates has minor deviation. The significance of differences is difficult to discern via inspection. However, small variances between the modelled values are magnified geometrically by the relationship between nucleotide and codon mutation frequencies, as per Additional File 2, and exponentially by the action of PCR amplification. Note similarities between p
ij
and for all estimates, where and denote the complements of nucleotides i and j, respectively.