Using DenseFly algorithm for cell searching on massive scRNA-seq datasets

BLEND: a fast, memory-efficient and accurate mechanism to find fuzzy seed matches in genome analysis

Generating the hash values of short subsequences, called seeds, enables quickly identifying similarities between genomic sequences by matching seeds with a single lookup of their hash values. However, these hash values can be used only for finding exact-matching seeds as the conventional hashing methods assign distinct hash values for different seeds, including highly similar seeds. Finding only exact-matching seeds causes either (i) increasing the use of the costly sequence alignment or (ii) limited sensitivity. We introduce BLEND, the first efficient and accurate mechanism that can identify both exact-matching and highly similar seeds with a single lookup of their hash values, called fuzzy seed matches. BLEND (i) utilizes a technique called SimHash, that can generate the same hash value for similar sets, and (ii) provides the proper mechanisms for using seeds as sets with the SimHash technique to find fuzzy seed matches efficiently. We show the benefits of BLEND when used in read overlapping and read mapping. For read overlapping, BLEND is faster by 2.4×-83.9× (on average 19.3×), has a lower memory footprint by 0.9×-14.1× (on average 3.8×), and finds higher quality overlaps leading to accurate de novo assemblies than the state-of-the-art tool, minimap2. For read mapping, BLEND is faster by 0.8×-4.1× (on average 1.7×) than minimap2. Source code is available at https://github.com/CMU-SAFARI/BLEND.

A Hashing-Based Framework for Enhancing Cluster Delineation of High-Dimensional Single-Cell Profiles

Although many methods have been developed to explore the function of cells by clustering high-dimensional (HD) single-cell omics data, the inconspicuously differential expressions of biomarkers of proteins or genes across all cells disturb the cell cluster delineation and downstream analysis. Here, we introduce a hashing-based framework to improve the delineation of cell clusters, which is based on the hypothesis that one variable with no significant differences can be decomposed into more diversely latent variables to distinguish cells. By projecting the original data into a sparse HD space, fly and densefly hashing preprocessing retain the local structure of data, and improve the cluster delineation of existing clustering methods, such as PhenoGraph. Moreover, the analyses on mass cytometry dataset show that our hashing-based framework manages to unveil new hidden heterogeneities in cell clusters. The proposed framework promotes the utilization of cell biomarkers and enriches the biological findings by introducing more latent variables.

Supplementary Information

The online version contains supplementary material available at 10.1007/s43657-022-00056-z.

Cellular Heterogeneity of the Heart

Recent advances in technology such as the introduction of high throughput multidimensional tools like single cell sequencing help to characterize the cellular composition of the human heart. The diversity of cell types that has been uncovered by such approaches is by far greater than ever expected before. Accurate identification of the cellular variety and dynamics will not only facilitate a much deeper understanding of cardiac physiology but also provide important insights into mechanisms underlying its pathological transformation. Distinct cellular patterns of cardiac cell clusters may allow differentiation between a healthy heart and a sick heart while potentially predicting future disease at much earlier stages than currently possible. These advances have already extensively improved and will ultimately revolutionize our knowledge of the mechanisms underlying cardiovascular disease as such. In this review, we will provide an overview of the cells present in the human and rodent heart as well as genes that may be used for their identification.