ImpulseDE2 troubleshooting

[av894]

Hi,

I am using ImpulseDE2 on a case-control dataset with 3 different timepoints. I have a few issues I'd like to solve and unfortunately few descriptions in the literature. Sorry in advance for all the questions, but I am not a code developer...The rationale of this package is quite elegant but unfortunately the literature is not very explicative.

1. I normalised my counts based on tpm and kept just counts > 0.5. I didn't normalise using log2 before starting the analysis. Is this a required step? I can't get my head around it by reading the papers.

2. When running runImpulseDE2(), for one of the trmt I have unequal numbers of case and control (e.g. 1 ctrl at 1h, 3 ctrl at 24h and 2 ctrl at 48h && 2 case at 1 h, 2 case at 24h and 2 case at 48h). If I run the function using batch (replicates) it stops, but if I remove it's fine. Is there a way to add this as a dispersion/confounding element?

3. When I get the results from runImpulseDE2(), there are lots of variables I am not familiar with. For my results, I am interested in looking at padj < 0.05 but what about the z-scores? I assume they relate to up and downregulation (then plotted in the heatmaps) but I can't seem to locate them in the LargeImpulseDE2 object.

4. From LargeImpulseDE2 object I get 2500 DEGs but vecDEGenes is full of NAs. Also, there is a discrepancy between this information and what I get from padj < 0.05, where I just get 12 DEGs.

5. When plotting the heatmaps I can either plot case, control or combined. What combined is showing? Is it plotting the trend of the case VS control? It's useful to have information about case and control separately during the timecourse but I also need to know how the trmt is changing expression compared to control.

6. Is there a way to plot the rownames with the heatmaps? Knowing gene expression trends without knowing which genes are actually expressed is not very useful.