Whole Genome de novo Assembly of Two Bayer Elite Rice Lines

We performed genome sequencing and de novo assembly for two elite indicarice lines that are parents for a proprietary commercial hybrid.  Initial assemblies were performed using ALLPATHS-LG on Illumina reads from paired-end and mate pair libraries.  Fosmid-end sequences and PacBio long reads were then used for further scaffolding and gap filling.  A genetic map constructed from sequencing data of 2000 F2 individuals was used to order and orient >300 scaffolds composing around 90% sequence length of each assembly.  Finally, remaining scaffolds were placed using the public Nipponbare genome as a reference.  The final assemblies comprise 1,244 and 1,522 scaffolds, with N50 scaffold sizes of 3.0 and 2.1 Mb and total sizes of 401 Mb and 404 Mb, respectively.  The iterative assembly allowed us to follow the progress with each added dataset and demonstrated the value of the mate pairs, long reads, and genetic map.

An assessment of gene content in the assemblies using CEGMA showed that 96% and 98% of eukaryotic conserved genes were at least partially covered in our two assemblies, respectively, compared to 98% in both the 93-11 and Nipponbare genome assemblies.  Repeat analysis revealed that around 9% repetitive sequence is missing from the assemblies of the two Bayer rice lines, accounting for their smaller sizes in comparison with the two public genomes.  The reference genome assemblies of these two elite lines provide a valuable resource for marker and gene discovery in our rice breeding program, as well as for reference-based assemblies of additional Bayer indica lines.