AnnZarro: scalable, interactive exploration of cell-by-cell and gene-by-gene relationships in single-cell data
Cell-by-cell and gene-by-gene relationships, such as kernels and correlations, shape most single-cell analyses but are rarely inspected directly due to a lack of visualization tools that display them. AnnZarro is a point-and-click browser viewer that shows these matrices linked to an AnnData object's cells x genes data. Using AnnZarro, experimental biologists can explore data without code and…
<AnnZarro> is a powerful tool designed for the interactive exploration of cell-by-cell and gene-by-gene relationships within single-cell data. Its primary function is to present complex matrices, which are often challenging to interpret due to the lack of visualization tools, in a user-friendly manner. With AnnZarro, both experimental biologists and computational biologists can easily delve into their data without needing to write any code or perform complex computations.
The browser viewer displays matrices linked to an AnnData object, which contains cell-by-cell and gene-by-gene data. Each panel of the viewer focuses on a single cell and a single gene, allowing for a clear and concise examination of the relationships between these elements. When a user clicks on a cell, the corresponding row of any cell-by-cell matrix is highlighted over an embedding or spatial coordinates.
Similarly, clicking on a gene highlights its row in any gene-by-gene matrix over a gene plot, such as a volcano plot. The focus of the viewer also reveals which genes change in the selected cell and which cells distinguish the selected gene.
One of the key strengths of AnnZarro is its ability to handle large datasets that exceed memory limitations on a typical laptop or shared server. The viewer only reads the matrix rows and columns displayed in each panel, enabling users to explore matrices that are far larger than what can be held in memory. When a large dataset is opened, AnnZarro presents a reproducible subset, and users can then navigate through the remaining cells in manageable parts.
For instance, a 100,000-cell subset of the Tahoe-100M dataset opens in just 4 seconds on a laptop and remains interactive.
Despite having the option to display all 95.6 million cells in a single plot, AnnZarro remains highly efficient. On a laptop, this process takes between 5 to 7 seconds, providing users with an interactive view of the entire dataset. The tutorial accompanying AnnZarro demonstrates its capabilities by preparing a dataset using Kompot for differential abundance and expression, a diffusion kernel, and gene correlations.
The dataset is then explored through six worked examples, showcasing the tool's flexibility and utility in various biological contexts.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
