Daniel H. Wagner Associates has developed algorithms
to improve assemblies of DNA sequences by local refinement. Given
a set of assembled primary sequences and the corresponding consensus
sequence, the algorithms use Markov Chain Monte Carlo (MCMC) modeling
to stochastically sample the distribution of assemblies that are
consistent with the primary sequences, under the assumption that
the original assembly is correct at large-scale. Further MCMC
modeling is used to sample consensus sequences that are consistent
with the sampled assemblies.
The local refinement model incorporates assumptions about the
likelihoods of substitution, insertion, and deletion errors in
the primary sequences. The algorithms will output statistical
information about the number and likelihood of local alternative
assemblies.
