Back to BACs?

Historical approaches for sequencing complex genomes used large insert BAC libraries (>150kbp inserts). BACs were end sequenced and fingerprinted, generating minimal tiling paths (MTPs) of BACs, which were then sequenced one BAC at time. Despite being both costly and time consuming it overcame many assembly problems associated with heterozygosity, repeats, and pseudogenes.

With the advent of next generation sequencing (NGS) de novosequencing of whole genomes with short reads became the method of choice. However, this approach struggles to resolve common repeats and homologous regions which are common in large, repeat rich polyploid genomes such as bread wheat.  Even in human good NGS assemblies N50 contig sizes are less than 10kb, or 24kb with added fosmid based sequences, whereas the human BAC based assembly’s N50 is 38.5Mb.

Here we detail a scalable, cost effective, high throughput pipeline to sequence BACs. BACs are grown and processed in 384 well format and a 48 x48 dual index barcode employed to facilitate pooling up to 2304 BACs in a single Illumina lane. Using this approach to sequence the minimal tile path of Barley chromosome 2H we observed the assembled sequence content per BAC averaging >140kbp with < 10 contigs per BAC and an N50 > 47kbp. Assemblies are then further improved by constructing mate pair libraries from 384 BAC pools and using this data to scaffold contigs.