Looking for Support with the correct vignette to follow / contrast setting

[Smeerlap]

Dear developers,

thank you for this amazing tool and the detailed vignettes. I am not a trained bioinformatician, but the vignettes are very easy to follow and I managed to re-create all the plots with my own dataset.

I have worked with scRNAseq data in the past. After reading through and following your vignettes, I am uncertain how to approach my own dataset and was hoping that you could guide me a little bit.

For my project, I am analyzing a mouse snRNAseq dataset with two conditions ("healthy" and "KO") with samples from 4 different timepoints (4, 6, 8 and 12 weeks). We are very interested in how the cellular communication in our disease model changes across time. So I am interested in finding out 1) How communication changes across time in the KO group. 2) If / how these time-dependent changes over time differ between the two groups.

For each condition-age pair (e.g. KO_4w or healthy_12w), I have 2 biological replicates, except for the healthy_8W and the healthy_6W pair, which only contain one biological replicate (therefore, I have a total of 14 samples).

In getting started, I am thinking about two aspects:

1. If I understand your vignettes correctly, I should follow the "sample-agnostic" workflow (https://github.com/saeyslab/multinichenetr/blob/main/vignettes/basic_analysis_steps_MISC_SACL.knit.md) since my groups all have < 4 biological replicates and pseudobulking might result in misleading results. As it states in the vignette, I cannot perform mutlifactorial experimental designs with this approach.

2. How do I best set my contrasts to address the biological questions that I have? Is it just not possible to address these questions with my setup (i.e. considering each condition-age pair as a group of interest) because of the limited sample size? Or is there a way to formulate the contrast design to circumvent these issues? Or would you recommend to group the samples more coarsely (e.g. "young" = weeks 4+6 and "old" = weeks 8+12) to increase sample size in each group and then follow the standard pseudobuling vignette?

Sorry for all the questions but as you can probably tell, I am getting a bit lost in all the details and my rather complex experimental setup.

Thanks a lot for your help!

All the best!