broadinstitute/single_cell_analysis

Documents used for workshops on single cell analysis

Introduction to Single-cell RNA-Seq Analysis

Summary: These are the materials for a single-cell RNA-Seq analysis workshop.

Included are:

• cut_and_paste.txt - You can cut/copy and paste code from this document into your RStudio console to work through the presentation. (On a windows machine word pad or RStudio worked best for opening this document.)

• url.txt - URL for on-line presentations

single_cell_analysis's Issues

Batch Effect Corrections

Dear Timothy,

I came across your workshop material, trying to find a way to correct for batch effects in scRNA-Seq data with Seurat. First of all: thanks for sharing!
But I do have some issues understanding how your batch correction approach works. If I get it right, you basically are extracting information about batch from the @data.info slot, binning batches 1 to 4 into one batch and 5 to 6 into the second batch, adding this to the Seurat object and then regressing for it. But how can the computer regress out batch effects without knowing what genes or other confounders are driving a potential batch effect? Could you elaborate on how this is achieved?

Adding to my confusion, when you are showing the TSNE Plot (https://rstudio-pubs-static.s3.amazonaws.com/245975_cd7f2637adea45db83873c5c1f421e07.html#/68), there is one cluster present that solely consists of cells originating from bipolar1-4 (i.e. batch1) and a second one right underneath that is solely made up of cells from bipolar 5 and 6 (i.e. batch2). Isn't this the kind of batch effect that was supposed to be removed?

Best,

Tobias