It is important to note that some details of the results may be sensitive to aligner algorithms and parameters. For example, a read whose best alignment has many errors may be left unmapped with one set of parameters, potentially contributing to coverage bias, or mapped with alternative parameters, potentially contributing to error bias. Similarly, parameters often determine whether a particular base is categorized as a mismatch, insertion, or deletion. As much as possible, we have addressed these issues by using aligners and parameters for each technology that have been validated as producing useful alignments for other applications. In our experience, the statistics we have presented are robust to reasonable substitutions of aligners and parameters.
