Twenty micrograms of gDNA was processed to create SMRTbell sequencing templates > 10 kb (average insert size 17 kb) and sequenced using a PacBio RS II System in which polymerase-MagBead-bound templates were loaded at an on-plate concentration of 150 pM. Templates were subsequently sequenced using DNA Sequencing Kit 2.0, with data collection of 180 mins (Pacific Biosciences). Genomes were assembled using HGAP (Chin et al. 2013 (link)) with default parameters in SMRT Analysis Suite version 2.1 (Pacific Biosciences). Additional manual assembly of contigs was carried out in cases of unique overlapping sequence. Consensus sequence polishing was done using the Quiver algorithm in Genomic Consensus version 0.7.0. Base modification analysis was performed by mapping SMRT sequencing reads to the respective assemblies using the BLASR mapper (Chaisson and Tesler 2012 (link)) and SMRT Analysis Suite version 2.1 using standard mapping protocols. Clustering of sequence motifs was performed using Motif Finder (https://github.com/PacificBiosciences/DevNet/wiki/Motiffinder). See Supplemental Methods for further details.