Recent years have seen remarkable innovation in sequencing technologies, making them high-throughput, faster and more cost effective. However, isolation of the target DNA to be sequenced has remained a bottleneck. Three new reports in Nature Methods provide an alternative to the traditional PCR- and BAC-based approaches to this step. Two of them describe microarray-based enrichment of target DNA, whereas the third involves targeted multiplex amplifications, all of which are ideally suited to large-scale, multiplex and cost-effective genome resequencing.

Okou et al. and Albert et al. have devised similar methods in which genomic DNA is fragmented, repaired by addition of adapters and captured on custom-made high-density microarrays. Following elution and amplification, this target pool is ready for resequencing. Both methods prove to be highly accurate — the method of Albert et al. captured 6,700 exons, adding up to 5 Mb, within which it identified 94–97% of variants, whereas Okou et al. achieved >99% identification in a contiguous 300-kb region on the X chromosome.

The approach of Porreca and Zhang et al. is quite different. In the first step, the authors synthesized a library of targeting oligonucleotide precursors directly on a microarray. Upon release from the array, the oligonucleotide precursors were PCR amplified and digested to create single-standed 'capture probes', the ends of which were designed to hybridize on either side of the target sequence. The 'capture' itself involves polymerase-driven gap filling and ligation to create circular DNA molecules that are then enriched, amplified and can be turned into a shotgun sequencing library.

This multiplexing approach is highly specific (∼98% of mapped reads corresponding to targeted regions) and can capture ∼10,000 targets in one reaction, raising the possibility that all annotated protein-coding human sequences could be queried in just a few reactions. However, the resequencing data at heterozygous positions are skewed away from the expected 0.5 ratio, something that warrants further investigation but might be due to inefficient hybridization to genomic DNA. Improving the uniformity of target capture and amplification is another important focus for future improvement — the ∼10,000 targets that were captured represent 20% of the total targeted sequence.

All three approaches bring important advances over the exisiting 'front-end' methods, in terms of throughput and cost. They are also free of the sequence-related constraints that plague restriction-enzyme-based approaches. Further improvements in these 'genome partioning' methods are undoubtedly just around the corner and, with them, cost-effective, high-throughput, targeted genomic resequencing.