Does Harmony batch correction for scRNA-seq require the same populations to be present in all samples?

Question

[Lucy]

Hi,

I would like to use Harmony to remove batch effects from my 10x Genomics scRNA-seq data combined from three donors. However, I am wondering whether Harmony requires the same cell populations to be present in all samples, or how it deals with a population that is e.g. unique to one donor?

Many thanks,

Lucy

[shoujun.gu]

looks like it requires same population.

Answer 1

[igor]

No. They illustrate that a point with the main figure on their tutorial page:

Harmony uses fuzzy clustering to assign each cell to multiple clusters, while a penalty term ensures that the diversity of datasets within each cluster is maximized.

Answer 2

[Lucy]

Thanks, this shows cell types that are present in two out of the three datasets. Does this work equally well if one of your populations is only present in a single dataset?