This is a test version of Biostars. For the public version, visit https://www.biostars.org.
How to plot LMER estimate values in heatmap ?

I am doing transcription factor activity analysis on my single cell data and I ve used lmer model for statistical comparison between my two condition.

I want to plot my results in heatmap ( TFs as rows, Celltype as columns) . I think I can use esiimate values generated from lmerTest::lmer() function as they are like fold change values but I am not sure.

Should I use estimates values on plots directly without scaling them ? If i use real estimates values ( max 0.3 , min - 1.5) , i get really weird plot.

cell scrnaseq tf lmer transcription
• 893 views • link
written by carolofharvest • carolofharvest

Some might know what "lmer model" for "transcriotion factor analysis" is. Why don't you show representative data so people can suggest visualization. Maybe a link to a paper doing similar?

• 0 views • link
written by ATpoint ATpoint

HI ,

This is the link of the plot (see figure K) ; https://www.nature.com/articles/s41590-024-01994-8/figures/14

Here they ve used log(medFC) , scores are calculated by using PROGENy. I ve noticed that estimate values and foldchange values can have different signs , so different direction in pathway activity; I suppose calculating foldchange is just a superficial way of showing direction compared to lmer estimates because lmer also takes into account random sample effect.

But I ve tried calculating foldchanges as they did in this article , but i dont know how they handled NANs produces after log transform (i ve tried adding pseudocount). And median scores for celltypes in different conditions are not always positive , cell types with negative scores shows downregulation of pathway , in other ways upregulation of genes that have negative weight in that pathway.

And this is the link of the article :

https://doi.org/10.1038/s41590-024-01994-8

• 0 views • link
written by carolofharvest • carolofharvest

0 answers

No answers yet.

Log in to answer this question.

More posts like this