Combining the outputs of multiple sequencing technologies might create a composite data set whose overall bias is reduced. Two technologies provide complementary coverage if, on the same sample, they tend to fill in each other's low-coverage regions. Complementary technology mixtures should have bias statistics that are better than either one of the components. Precedent for this approach stretches back to the practice of combining data from dye-terminator and dye-primer chemistries in Sanger sequencing to reduce error biases [33]. Note that there can be other benefits from mixing technologies, by taking advantage of a broader range of complementary properties (and not just bias). For example, for genome assembly there are benefits from combining the long, relatively unbiased but lower accuracy reads from Pacific Biosciences with shorter Illumina reads that provide per-base accuracy [34-36].
