How we could create reference genomes for many horse breeds at about $30K each

To understand the genomes of a population, a typical strategy has been to create a reference sequence for one individual, then sequence other individuals, and align their reads back to the one reference sequence, to define variants with it. This approach has great utility and yet is limited. For example sequences present in one individual but not the reference are likely to be missed.

Enabled by recent technological improvements, we describe here an approach for affordably sequencing and assembling the genomes of multiple individuals, to very high quality. Horses would provide an excellent test case.

Our approach has short-range, medium-range and long-range components. The short-range component is based on 2x250 base Illumina reads from a single PCR-free library. These data are straightforward to generate and inexpensive. We developed an algorithm DISCOVAR de novo to rapidly assemble them. Indeed library construction, sequencing and assembly could be executed in a week.

For a human genome, we obtained 125 kb contigs. The assemblies represent both haplotypes, and perfectly (up to defects every ~50 kb, typically in long homopolymers). For rhinoceros we obtained contigs of size 117 kb.

As a medium-range component,  we will use data from a jumping or related-type library, again sequenced by Illumina, which will scaffold contigs and fill in between them. Finally to obtain long-range structure, we will use BioNano Genomics maps. Optimal integration of these data types into a turnkey method will require substantial R&D, however an initial demonstration for horses could be carried out this year.