geosketch: Geometric Sketching

In RuiyuRayWang/scWidgets: A collection of scRNAseq analysis functions

Description

Geometric Sketching to subsample data from single cell genomics into smaller representative subset of cells, while preserving biological complexity, highlighting rare cell states, and accelerating complex analysis such as data integration.

Usage

geosketch(object, sketch.size = 1000, assay = "RNA", slot = "data", k = 20)

Arguments

object	A Seurat object.
sketch.size	Number of cells to sketch. By default sketch 1000 cells.
assay	Assay used for sketching. "RNA" by default.
slot	By default slot = "data".
k	Target rank of the low-rank SVD. k should satisfy k << min(m,n). By default k = 20.

Details

The original package 'geosketch' is written in Python. With 'reticulate', we implement it in R and provides a function that directly works with the Seurat object.

Publication: Hie, B., Cho, H., DeMeo, B., Bryson, B., and Berger, B. (2019). Geometric Sketching Compactly Summarizes the Single-Cell Transcriptomic Landscape. Cell Systems.

Github repo: https://github.com/brianhie/geosketch

Value

A sketched Seurat object.

Examples

## If haven't installed already.
reticulate::py_install("geosketch")

## mgc has 7527 cells.
mgc.sketched <- geosketch(mgc)

RuiyuRayWang/scWidgets documentation built on Dec. 18, 2021, 11:54 a.m.