Bayesian Genome Assembly and Assessment by Markov Chain Monte Carlo Sampling

摘要: Most genome assemblers construct point estimates, choosing only a single sequence from among many alternative hypotheses that are supported by the data. We present Markov chain Monte Carlo approach to assembly instead generates distributions of with posterior probabilities, providing an explicit statistical framework for evaluating and assessing uncertainty. implement this in prototype assembler, called Genome Assembly Bayesian Inference (GABI), illustrate its application bacteriophage X174. Our sampling strategy achieves both good mixing convergence on Illumina test data X174, demonstrating feasibility our approach. summarize distribution generated GABI as majority-rule consensus assembly. Then we compare external assemblies same data, annotate those assigning probabilities features common GABI’s graph. is freely available under GPL license https://bitbucket.org/mhowison/gabi.