Szalay, T. & Golovchenko, J.A. Nat. Biotechnol. 33, 1087–1091 (2015).

Nanopore sequencing holds the potential to deliver long, single-molecule reads that should be ideally suited for de novo genome assembly. But the current accuracy of this technology—only ∼85%—makes this application challenging. To correct many of the errors in nanopore reads, Szalay and Golovchenko developed PoreSeq, an algorithm that increases accuracy to up to 99%, as shown for a bacteriophage genome. PoreSeq models uncertainties in the measurement of changes in ion current as a DNA molecule is ratcheted through the pore and compares multiple reads of the same region to compute the most likely sequence for each read. The authors used these error-corrected reads to assemble the genomes of Escherichia coli and the λ-phage. They also were able to call single-nucleotide variants with 99.1% accuracy at 16× coverage, an order of magnitude lower than previous results.