scLKME: Landmark-based Multi-sample Single-cell Data Analysis

scLKME is an approach for sample-level analysis of multi-sample single-cell data. It uses landmark-based kernel mean embedding to generate sample embeddings. scLKME includes two steps:

  1. cell sketching: identify a subset of cells as landmarks to summarize the cell landscape across samples.

  2. kernel mean embedding: transform cell distributions using kernel mean embedding and align them at the landmarks.

The workflow of scLKME is as follows:

scLKME workflow figure

Get started

  • To install the sclkme, see the Installation.

  • For api usage, see the API.

  • For data analysis, check out the Examples.