Comparing Analysis Strategies for scRNA-seq Data: Separate vs. Merged Analysis of Spleen Samples from Different Conditions

Question

[Shukai]

I have 16 scRNA-seq spleen samples from mice, divided into four conditions with four replicates each. All samples were prepared and sequenced together, suggesting minimal batch effects.

I need to compare cell type proportions and perform DEA between conditions. I'm considering two approaches:

Separate Analysis: Process each condition separately with Seurat. Merged Analysis: Merge/integrate all samples and analyze together.

(By Analysis I am referring to QC at both cell and gene level, followed by the standard Seurat pipeline with normalization, clustering, and cell annotation by looking at marker expressions.)

I tried both approaches and lean towards the merged analysis. Preliminary UMAP results from the merged Seurat object of all 16 samples without integration (before QC) show no dramatic clustering by condition.

Questions:

1. Should I analyze my samples together or separately for each condition?
2. If analyzing together, is integration necessary given the uniform processing and minimal clustering by condition in the UMAP?

Any potential pitfalls with either strategy?

[Ram]

1. You can edit your post and add relevant details. There is no need to add separate comments with relevant details especially when it's not in response to a question/request by someone else.
2. Please use proper formatting - see your post now and observe how your list of questions and your code look. Without proper formatting, a post looks like a wall of text and no one wants to read a wall of text.

[Shukai]

Thank you Ram for the comments! Noted for future posts.