what is residualModelFormulaStr in monocle?

Question

[John]

Hi, I want to Cluster cells using marker genes, I use monocle, I get the following error, Here I don't know what to replace with ~Media, can anyone please explain what is residualModelFormulaStr ?what should I give there?

> marker_diff <- markerDiffTable(HSMM[expressed_genes,], cth, residualModelFormulaStr = "~Media + num_genes_expressed", cores = 1)
Error in `[.data.frame`(pd, , i) : undefined columns selected

Answer 1

[Batu]

Hello, I also could not find what it is exactly. But I solved it by trial and error, it is solved when i delete Media, leave num_genes_expressed only.

marker_diff <- markerDiffTable(pfcm[expressed_genes,], cth, residualModelFormulaStr = "~num_genes_expressed", cores = 2)

But I also need to learn what it is, since it is used in differentialGeneTest function.

[davidjoule94]

I think you're right. Here is how they set up the experiment: To initiate myoblast differentiation, we switch media from a high-mitogen growth medium (GM) to a low-mitogen differentiation medium (DM). So they need to exclude the effects of "culture media" from clustering. If you don't have such effects that potentially cause bias then there is no need to add it.

[davidjoule94]

I think you're right. Here is how they set up the experiment: To initiate myoblast differentiation, we switch media from a high-mitogen growth medium (GM) to a low-mitogen differentiation medium (DM). So they need to exclude the effects of "culture media" from clustering. If you don't have such effects that potentially cause bias then there is no need to add it.

Answer 2

[Batu]

In this issue in GitHub, you can find the andwer about formula string: https://github.com/cole-trapnell-lab/monocle-release/issues/151