Figure 1.

Schematic of smMIP method. (A) Molecular inversion probes (MIPs) consisting of two 16–24 nt “targeting arms” (dark gray) joined by a constant 28-nt “backbone” sequence (light gray) and a 12-nt degenerate “molecular tag” (red) were designed for the coding exons (light-blue rectangle) of 33 cancer-related genes. Targeting arms were complementary to sequences flanking individual regions of interest, each 112 nt in length. (B) Probes are pooled, hybridized to genomic DNA, and polymerase and ligase were added to “gap-fill” the reverse complement of the genomic DNA to which the probe is hybridized (light-blue) and ligate the probe into a single-stranded circle. (C) After exonuclease treatment and PCR, sequencing library molecules consist of platform compatibility (black), probe backbone (light gray), targeting arm (dark gray), copied target (light blue), molecular tag (red), and sample-specific index introduced during PCR (green). Massively parallel sequencing is used to collect three reads (dark blue). (D) Overlapping read-pairs are reconciled to form “fr-reads” (dark blue), assigned to samples via the sample-specific index sequence (green) and individual capture events via the molecular tag (red). (E) Groups of fr-reads assigned to the same probe via alignment to the reference genome and sharing the same molecular tag and sample index form a “tag-defined read group” (TDRG). Random errors (yellow) that occur during library construction and sequencing may be present in some members of the TDRG at some positions. (F) TDRGs are used to call a single molecule consensus sequence (“smc-read”) for the captured target sequence that is robust to such errors.