Abstract
Sequencing-based microscopy is an emerging family of techniques for locating the positions of molecules in a tissue or biological sample with-out traditional optics. Instead, spatial information is gathered through a network of DNA sequences that tag individual molecules. Such information is retrieved using high throughput sequencing technology and reconstructed computationally into an image. Multiple proposed chemistries and proof of concept experiments have established the feasibility of this approach on simple model systems. However, methods are needed to assess the validity of reconstructed images in the absence of ground truth knowledge or parallel use of optical techniques. To address this, we identified a set of ground truth-agnostic properties, spatial coherence measurements, that may be computed in simulated or experimental sequencing-based microscopy data. Spatial coherence represents a network’s potential to preserve spatial relationships through its topology. The measures are based on a generalization of Euclidean geometry to spatial networks, and they indicate when such networks deviate from Euclidean laws. These deviations, e.g. due to the presence of non-spatially correlated connections, create contradictory constraints that lead to distortions in image reconstruction. We propose that spatial coherence may be employed as a generic metric of overall quality of spatial information in sequencing-based microscopy even in the absence of secondary validation, as it is based on a fundamental geometric criteria.
Competing Interest Statement
The authors have declared no competing interest.
