Perfect Long Read Assembly and the Rise of Pan-Genomics

New single molecule sequencing technologies from both Pacific Biosciences and Oxford Nanopore are producing extremely long reads, commonly over 10,000 bp and some approaching 100,000 bp. These data have the potential to dramatically improve de novo genome assembly, structural variation discovery, transcript analysis, and many other important genomic assays. The main limiting factor for applying these technologies is their high error rates, typically 10% to 15% for PacBio SMRT sequencing, and even higher for Oxford Nanopore using their early access instruments.

Here we present our work examining both PacBio and Oxford Nanopore single molecule sequencing, comparing their sample requirements, read length & coverage distributions, and error models. We also present the results assembling these data on their own using “self-correction” techniques where the long reads are aligned and error corrected relative to each other, and using “hybrid” approaches error correcting the long reads with high coverage short read sequencing from Illumina. Using these approaches, we have been able to achieve near perfect assemblies using both Oxford Nanopore and PacBio sequencing in small microbial genomes and of yeast (S. cerevisiae). With PacBio sequencing we have also achieved extremely high quality assemblies of several larger plant and animal genomes including of O. sativa (rice). From this assembly, we have isolated thousands of regions specific to the rice strain Indica not present in Nipponbare along with several megabases of copy number variation that were previously unresolved. Many of the most significant differences contain genes and other loci associated with agriculturally important traits.