In SydneyBioX/scMerge: scMerge: Merging multiple batches of scRNA-seq data

Introduction

This is a benchmarking single-cell data comprising of 3 different human lung adenocarcinoma cell lines. This experiment was designed induce different batch effects into the data.

Integration challenge

Prior to integration, there is a strong separation effect by multiple batch effects.
One of the batch effects is the different protocols with significant depth differences.
The experimental design poses extra difficulties as cell lines were cultured separatedly and the same batch was processed in 3 different ways.

Data description

Data source:

| Type of merge | | Name | | ID | Author | DOI or URL | Protocol | Organism | Tissue | # of cell types | # of cells | # of batches | |---------------|---|-----------|---|-----------|--------|--------------------------------------------|----------------------------------|----------|---------------------------|-----------------|------------|------------------| | Across platforms with significant depth difference | | CellBench | | CellBench | | https://github.com/LuyiTian/CellBench_data | Cel-seq2, Drop-seq, 10x Chromium | Human | Adenocarcinoma cell lines | 3 | 1401 | 3 per cell types |

Relation to the scMerge article: Supplementary Figure 12.

Data visualisation

tSNE plots by cell types and batch

{width=100%}

Integrated `scMerge` data

Data availability: CellBench Data (in RData format)
scMerge parameters for integration:
- Unsupervised scMerge
- kmeans K = (3,3,3)
- Negative controls are human scSEG