Posters - Schedules

Posters Home

Spatial Transcriptomics (ST) is a powerful Next Generation Sequencing-based technique that combines expression information and spatial context for a given sample, and thus allows new insights into the cellular composition and spatial organization of tissues. Several ST techniques with different structures exist, e.g. grid-based spot data, or marker-based regions of interest (ROI), but most current approaches are limited to multi-cell resolution. A precise estimation of the abundance of cell types in a chosen region of interest is therefore essential for accurate ST data interpretation.
We compared the performance of deconvolution algorithms regarding the inference of immune cell types on original and published NanoString GeoMx ST data from multiple sclerosis (MS) lesions and lung tumor tissue. Both published and single cell-based custom immune cell profiles were used for the cell type deconvolution, and effects of raw data quality and algorithm parameter settings on the estimated immune cell populations were assessed. With preprocessed immune signature matrices as basis for computational deconvolution, SpatialDecon, FARDEEP, and EPIC consistently identified core immune cell populations in a subset of seven MS regions of interest. Detection of other cell types was more variable, and algorithm-dependent. Lower sample quality generally impeded accurate cell type deconvolution.

The coronavirus disease (COVID-19) can alter leukocyte phenotype in terms of the disease severity, including neutrophil activation signatures in severe cases. In recent years, accumulating evidence has revealed unexpected phenotypic heterogeneity and diverse functions of neutrophils in many other diseases. However, the complexity of neutrophil phenotypes and their relative impacts on COVID-19 pathogenesis have not been well addressed. Here, we integrated public single-cell and bulk RNA-sequencing data from healthy donors and COVID-19 patients to investigate neutrophil heterogeneity and uncover how they contribute to disease pathogenesis. We identified and described neutrophil phenotypes with different activation signatures and enriched pathways. One of the phenotypes was associated with severe and fatal patients. Cell-cell communication analysis revealed different interactions among the neutrophil phenotypes. In addition, the bulk RNA-seq datasets analysis validated the relative abundances of neutrophils and fraction expansion of the specific phenotypes in severe COVID-19 patients. Our work provides a framework to understand the functional heterogeneity of neutrophils and sheds light on the prevention and treatment of COVID-19.

As the availability of biological sequencing data continues to grow at an exponential rate, efficient methods for storing, indexing, and analyzing this data have become increasingly important. Annotated de Bruijn graphs have emerged as a popular method for representing large sets of sequencing data, as they enable efficient storage of k-mer sets and their annotations in a compressed form. In turn, the Counting de Bruijn graph was developed as a generalization of the annotated de Bruijn graph, which allows supplementing each node-label relation with one or more attributes such as k-mer count or coordinates. The concept of the Counting de Bruijn graphs is utilized in the MetaGraph framework, which offers a unique approach to indexing global coordinates by utilizing a number of compression techniques for both the graph and the annotations. In this work, we present an algorithm for extracting read sequences from a Counting de Bruijn graph, which was implemented within the MetaGraph framework. This task is of critical importance as getting the read sequences of interest from which the graph was built would open new opportunities for downstream analysis after sequence search.