Can a New Microscopy Platform Help to Improve Clinical Outcomes?

Next-generation sequencing (NGS), also referred to as ‘second-generation sequencing’, ushered in an era of plummeting sequencing costs that has enabled rapid progress in the research of genetic diversity, variants, transcription and epigenetics, among other fi elds. In the clinical sphere, NGS spurred the dream of precision medicine (PM) or personalized medicine, in which the genome of each patient (or pre-patient) is sequenced, and an optimal treatment is designed based on one’s genetic code, lifestyle and environment [1] . Despite these gains and hopes, NGS is limited by its short-read approach and per-patient cost [2] , and PM has yet to exhibit transformative reach, with only a small proportion of US patients currently receiving genome-informed therapy [3] . In order for PM to become a standard of care, there is an urgent need for fast, cheap and reliable genetic diagnostics to complement existing techniques. We offer a novel high-speed atomic force microscopy ‘nanomapping’ platform [4] to help meet the challenge. NGS involves the sequencing and assembly of short reads (typically ~100 bp), which limits its capacity to map elements spanning longer regions of the chromosome – including repeats, structural variants, and heterozygous variants that differ between the two parent chromosomes. Given that these elements account for the majority of diversity and pathogenetic variation between individuals [5] , alternatives and modifi cations to NGS have been intensely explored. Single-molecule, ‘third-generation’ sequencers (e.g., PacBio and Oxford Nanopore) can produce reads of 10 kb or more, enabling the sequencing of previously inaccessible portions of the genome despite stochastic error rates of about 15%. Between these third-generation approaches, NGS improvements, and hybrid secondand third-generation assemblies, extremely accurate wholegenome or targeted sequencing efforts are now possible [2] . The effective translation of these techniques into a clinical setting is hindered by their time and cost, however: a whole-genome NGS run is, at best, currently priced at multiple thousands of US dollars, without considering computing or instrument costs, third-generation techniques are signifi cantly more expensive, and both require several days and substantial computing power. These costs are undoubtedly justifi ed in many cases, especially when a population-level genomic reference is being established [1] , but cheaper, faster, targeted alternatives are surely preferable in many other scenarios. For example, if PM is to confront the common and daunting challenge of heterogeneous, evolving cancer [6] , multiple rounds of rapid sequencing will surely be necessary to gather an updated diagnosis and gauge the effi cacy of a given treatment, causing NGS or single-molecule sequencing costs to escalate rapidly. To identify pathogenetic variants in a more targeted and effi cient manner, genetic mapping offers a potential alternative to sequencing. At signifi cantly lower cost and testing time, optical mapping techniques (e.g., Opgen and BioNano) can detect the locations of fl uorescent probes along extended strands up to 250 kb, and then compare the resulting map against a reference to determine the presence and character of indels, repeats or other variants [7] . Optical mapping has proved an important complement to sequencing that can essentially serve as a more detailed version of fl uorescent in situ hybridization (FISH), but it suffers from limitations of its own: the template is only sparsely sampled, the nicking required to introduce fl uorescent biomarkers can make heavily labeled portions prone to breakage, and most signifi cantly, experimental and optical limitations typically render variants shorter than 10 kb inaccessible to mapping [2] . We recently showed that genetic mapping need not compromise resolution for speed and affordability [4] . Instead of measuring strand length and mapping fl uorescent probes using optical techniques, which are realistically unable to resolve two probes closer than ~1.5 kb or longer [7] , we employ high-speed atomic force microscopy (HS-AFM) to image labeled DNA with effective resolution of ~15 bp. AFM was invented in the 1980s as a force sensing and imaging technique with resolution far surpassing that of optical microscopy [8] . The fundamental principle of AFM is similar to that of a record player: a sharpened tip detects topography as it is passed across a surface. The tip is attached to a fl attened cantilever with a laser refl ected off its rear face, such that a protrusion or depression encountered by the tip causes the cantilever to defl ect and the laser’s refl ected angle to deviate. The laser deviation, and thus the tip height, can be measured with nanoscale accuracy. With AFM tips routinely manufactured by chemical etching to achieve diameters less than 5 nm and tip translation across the surface driven by piezoelectric stacks capable of subnanometer displacement, a three-dimensional map with nanoscale accuracy is collected when the tip is rastered across a surface. An AFM can operate at ambient temperature and pressure or in liquid, and has been successfully employed to investigate a range of biological processes [9] . However, 65 2018