P0278
Towards Building Complete Genome Assemblies using BioNano Next-Generation Mapping Technology

Andy Wing Chun Pang , BioNano Genomics, San Diego, CA
Thomas Anantharaman , BioNano Genomics, San Diego, CA
Palak Sheth , BioNano Genomics, San Diego, CA
Xiang Zhou , BioNano Genomics, San Diego, CA
Jian Wang , BioNano Genomics, San Diego, CA
Joyce Lee , BioNano Genomics, San Diego, CA
Alex Hastie , BioNano Genomics, San Diego, CA
Zeljko J. Dzakula , BioNano Genomics, San Diego, CA
Han Cao , BioNano Genomics, San Diego, CA
High-quality assemblies are important when trying to understand the biology of genomes. Current short-read assemblers are memory intensive and have difficulties constructing contiguous assemblies; collecting deep coverage data by long-read technologies can be time-consuming and expensive. BioNano’s next-generation mapping data complements short-read data and alleviates the need to collect high-coverage long-read data.

BioNano Genomics Irys® System utilizes high-molecular-weight DNA to construct ultra-long genome maps. These assemblies reveal large structural variants and combined with sequencing assemblies produce hybrid scaffolds of unprecedented lengths, some spanning chromosomal arms. Chimeric joins are formed when short reads, molecules, or paired-end inserts are unable to span across long DNA repeats. These errors appear as conflicting junctions when one compares the sequence and the BioNano assemblies.

We developed a method to detect and resolve these conflicting junctions. At a conflicting region, where the BioNano genome map does not have long BioNano molecule support, the map is split into halves, thus resolving the conflicting junctions. If it does have BioNano molecule support, the sequence fragment is split. Using this approach on a karyotypically normal human genome, we resolved 12 (85.7%) erroneous translocation calls from the sequence assembly and one (11.1%) erroneous translocation call from the BioNano assembly. Then, by combining the two refined assemblies, we built long hybrid scaffolds achieving an N50 of 21.3 Mb (which is over 2.2 times the input sequence assembly N50).

Our plan is to use this novel hybrid scaffold functionality for constructing correct and contiguous reference assemblies for complex plants and animal genomes.