High-resolution human genome structure by single-molecule analysis

Abstract

Variation in genome structure is an important source of human genetic polymorphism: It affects a large proportion of the genome and has a variety of phenotypic consequences relevant to health and disease. In spite of this, human genome structure variation is incompletely characterized due to a lack of approaches for discovering a broad range of structural variants in a global, comprehensive fashion. We addressed this gap with Optical Mapping, a high-throughput, high-resolution single-molecule system for studying genome structure. We used Optical Mapping to create genome-wide restriction maps of a complete hydatidiform mole and three lymphoblast-derived cell lines, and we validated the approach by demonstrating a strong concordance with existing methods. We also describe thousands of new variants with sizes ranging from kb to Mb.

Fig. 1.

An overview of the Optical Mapping platform. Bulk microscope cover glass is cleaned with a strong acid, then treated with a silane mixture to make positively charged Optical Mapping surfaces (i). A silicon wafer is patterned with standard photolithography techniques, and then replicated into a flexible PDMS microfluidic device (ii) using soft lithography. Finally, pure, high molecular-weight DNA (iii) is isolated from cultured eukaryotic cells using a gentle detergent-based lysis protocol. The microfluidic device is adhered to the Optical Mapping surface, and the DNA solution is pumped through the microchannels, wherein the DNA is elongated and attached to the Optical Mapping surface via electrostatic interaction (iv). The DNA is incubated with a restriction endonuclease (v), which cleaves the DNA at its cognate sites. The cleaved DNA is stained and imaged on an epifluorescence microscope (vi) illuminated by an argon-ion laser (vii) and controlled by a computer workstation (viii).

Fig. 2.

An overview of the map assembly pipeline. Reference maps are generated in silico from the NCBI Build 35 human genome reference sequence (40), and used to seed an iterative process of pairwise alignment (which clusters together similar single-molecule maps) and local assembly (which generates a consensus optical map from a cluster of single-molecule maps). After several iterations of alignment and assembly, the consensus maps are aligned back to the reference map and analyzed for places where the consensus map differs significantly from the reference, indicating potential polymorphisms.

Fig. 3.

A representation of the structural variation found in four genomes analyzed by Optical Mapping. Variants from the CHM genome are depicted in green; GM15510 in blue; GM10860 in red; and GM18994 in gray. The inset depicts five example differences from the genome of GM10860: an extra cut, a missing cut, a 250 kb deletion, a 150 kb insertion, and a 150 kb inversion.

Fig. 4.

The optical map complements hybridization-based approaches. The optical map reveals that the gain in sequence detected by the Affymetrix SNP 6.0 platform (shaded region) is due to an inverted tandem duplication at this locus (red arrows).

Fig. 5.

A large insertion from GM15510 chromosome 7. Optical Mapping indicates a 90 kb insertion, confirming the large insertion that was indicated by a cluster of singleton fosmids reported by Kidd et al. (10) (red arrows). Included below the map is a montage of several of the single-molecule images that give evidence to support this insertion.

Fig. 6.

(A) Optical Mapping has greater ability to discern variation in repeat-rich regions than hybridization-based technologies. The first bar in each section is a genome-wide representation of variants discerned only by Optical Mapping (green), only by an alternate technology (blue for PEM, red for CGH), or by both technologies (gray). (For example, in the first bar of the PEM comparison, 76% of the variants were found only by Optical Mapping, 17% were found only by PEM, and 7% were found by both technologies.) Subsequent bars represent the same proportions, but include only variants that intersect with various classes of repeat. The proportions are substantially the same when comparing Optical Mapping to PEM, but Optical Mapping detects a greater proportion of variants intersecting repeats when compared to hybridization-based technologies (χ² test, p < 10^-7). (B) Optical Mapping-discerned variants are more evenly distributed between insertions (median size, 4.5 kb) and deletions (median size, 4.3 kb). We compared the sizes of indels discovered with Optical Mapping to platforms based on end-sequencing and hybridization. Indel size density was estimated for each dataset using a Gaussian kernel with a bandwidth of 0.3. Negative sizes represent deletions, while positive sizes are insertions. The Optical Mapping indels are more evenly distributed between insertions and deletions, perhaps due to the platform’s unique ability to detect large novel insertions.