Alignments of chloroplast noncoding regions for context-dependency analysis

Substitutions between closely related noncoding chloroplast DNA sequences
are studied with respect to the composition of the three bases on each
side of the substitution, that is the hexanucleotide context. There is
about 100-fold variation in rate, amongst the contexts, particularly on
substitutions of A and T. Rate heterogeneity of transitions differs from
that of transversions, resulting in a more than 200-fold variation in the
transitions:transversion bias. The data are consistent with a CpG effect,
and it is shown that both the A+T content and the arrangement of
purines/pyrimidines along the same DNA strand are correlated with rate
variation. Expected equilibrium A+T content ranges from 36.4% to 82.8%
across contexts, while G-C skew ranges from -77.4 to 72.2 and A-T skew
from -63.9 to 68.2. The predicted equilibria are associated with specific
features of the content of the hexanucleotide context, and also show close
agreement with the observed context-dependent compositions. Finally, by
controlling for the content of nucleotides closer to the substitution site
it is shown that both the third and fourth nucleotide removed on each side
of the substitution directly influence substitution dynamics at that site.
Overall, the results demonstrate that noncoding sites in different
contexts are evolving along very different evolutionary trajectories and
that substitution dynamics are far more complex than typically assumed.
This has important implications for a number of types of sequence
analysis, particularly analyses of natural selection, and the
context-dependent substitution matrices developed here can be applied in
future analyses.