Optimizations of Physical Genome Map Contiguity by In Silico Ligation

Genome assemblies based solely on short-read technologies are often fragmented due to structural complexities like tandem and interspersed repetitive segments, long-range structural variations, and dispersed segmental duplications. In diseases such as prostate cancer, it has been estimated that about 50% of all prostate cancers have recurrent gene fusions, structural variation events not observable with NGS¹. Algorithms that improve automation, reduce manual curation time, and improve data quality and confidence are sorely needed.

The BioNano Genomics Irys® System linearizes and molecularly barcodes long DNA molecules, yielding single molecule information contiguous up to megabase lengths. The single molecule information is assembled into genome maps; once compiled, the Irys System uses this information to scaffold the sequence assemblies, validate the accuracy of the sequences, and anchor the adjacent sequences in the proper order and orientation. The long genome map contiguities are further improved by bridging the breaks at fragile sites: fragile site breaks are breaks that occur when modified restriction enzymes introduce nick sites too close in proximity to each other and an unintentional double-stranded break in the DNA molecule occurs.

We present a computational pipeline (fragileSiteRepair) that improves structural variation detection sensitivity; automating the process of fragile site prediction, scaffolding of genome maps across fragile sites, and applying confidence scores based on single molecule alignments. The resulting fragile site repaired genome maps are highly contiguous with 25-100% increase in overall N50s with a corresponding increase in structural variation sensitivity.