This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Seurat integration join split

I have a crossover study design with 4 individuals with 4 treatments each for a single-cell experiment.

E.g.,

Person 1 - Treatment A, B C, D

Person 2 - Treatment A, B C, D

Person 3 - Treatment A, B C, D

Person 4 - Treatment A, B C, D

I load all 16 samples, run QC, merge, and joinlayers(). From here, I want to split the layers to run SCTransform and then integratelayers(), but am unsure whether I should split by "Person" (4 layers), "Treatment" (4 layers), or "Sample" (32 layers). My end goal is to create a UMAP where I can take a cluster to subset for downstream analysis, but that subsetted cluster needs to evenly represent treatment and person so the derived expression is not skewed towards a certain treatment or subject from which the cells originate from.

seurat single-cell sctransform integrated-analysis
• 254 views • link
written by michaelhojungyoon • michaelhojungyoon

First, you should perform the integration using each patient as a separate layer, as you mentioned, to generate a fully integrated UMAP. From there, you can subset the cells or clusters of interest, extract the count matrix, and then integrate again.

• 0 views • link
written by carolofharvest • carolofharvest

0 answers

No answers yet.

Log in to answer this question.

More posts like this