Significantly improved multiplex padlock capturing and large scale sequencing reveal hypermutable CpG variations

To take full advantage of the power of next generation sequencing requires large scale multiplex capturing and amplification of genomic regions of interest in multiple samples. We improved a previously developed padlock probe-based approach by 10,000-fold and evaluated it by identifying genetic variations in hypermutable CpG rich regions, tiling CpGs across human chromosome 21 with 53,777 probes in an unbiased manner. Two to three million mapped sequences derived from a single Illumina Genome Analyzer lane enable observation of 90.8%-94.0% of target sites with at least one read. Although the sites were not uniformly captured, ~85% and ~55% of all targets fell within a 100- and 10-fold range of each other, respectively . A total of 442937 genotypes were computed with confidence across six subjects and examined for single nucleotide polymorphisms and were found to be in between 98.4% and 100% agreement with independently obtained genotype data. We detected as many as 502 sites of variation not present in dbSNP, including 362 in targeted CpG locations not among the 2640 in dbSNP. CpG->CpA and CpG->TpG variations were found to be ~12.5x more abundant than non-CpG variations. Variation rates differed among subjects in a possibly ancestry-related manner. Our success suggests that genotyping hypermutable CpGs may be an efficient way of identifying common and rare mutations in human diseases and that padlock-probe based sequencing can reveal important aspects of human variation generally. We identified areas for further improvement and study.