Gigante, S. A. | Search bioRxiv

bioRxiv2021

Topological analysis of single-cell data reveals shared glial landscape of macular degeneration and neurodegenerative diseases

1One Sentence SummaryA novel topological machine learning approach applied to single-nucleus RNA sequencing from human retinas with age-related macular degeneration identifies interacting disease phase-specific glial activation states shared with Alzheimers disease and multiple sclerosis. 2 AbstractNeurodegeneration occurs in a wide range of diseases, including age-related macular degeneration (AMD), Alzheimers disease (AD), and multiple sclerosis (MS), each with distinct inciting events. To determine whether glial transcriptional states are shared across phases of degeneration, we sequenced 50,498 nuclei from the retinas of seven AMD patients and six healthy controls, generating the first single-cell transcriptomic atlas of AMD. We identified groupings of cells implicated in disease pathogenesis by applying a novel topologically-inspired machine learning approach called diffusion condensation. By calculating diffusion homology features and performing persistence analysis, diffusion condensation identified activated glial states enriched in the early phases of AMD, AD, and MS as well as an AMD-specific proangiogenic astrocyte state promoting pathogenic neovascularization in advanced AMD. Finally, by mapping the expression of disease-associated genes to glial states, we identified key signaling interactions creating hypotheses for therapeutic intervention. Our topological analysis identified an integrated disease-phase specific glial landscape that is shared across neurodegenerative conditions affecting the central nervous system.

neuroscience↗

bioRxiv2019

Enhancing experimental signals in single-cell RNA-sequencing data using graph signal processing

Current methods for comparing scRNA-seq datasets collected in multiple conditions focus on discrete regions of the transcriptional state space, such as clusters of cells. Here, we quantify the effects of perturbations at the single-cell level using a continuous measure of the effect of a perturbation across the transcriptomic space. We describe this space as a manifold and develop a relative likelihood estimate of observing each cell in each of the experimental conditions using graph signal processing. This likelihood estimate can be used to identify cell populations specifically affected by a perturbation. We also develop vertex frequency clustering to extract populations of affected cells at the level of granularity that matches the perturbation response. The accuracy of our algorithm to identify clusters of cells that are enriched or depleted in each condition is on average 57% higher than the next best-performing algorithm tested. Gene signatures derived from these clusters are more accurate compared to six alternative algorithms in ground-truth comparisons.

bioinformatics↗