**0**wrote:

Hi there,

I'm struggling a little bit with understanding how to interpret the heatmap output from the fission test data here:

First, I would like to just confirm that I understand the differences in the Wald and LRT padj values.

This heatmap (shown below) was generated by gathering the top 20 genes with the lowest padj values in the

`results`

from the whole time series. That is, these are the genes that at one or more time points showed differences between the two strains tests.However, if we want to know the p-values/padj for a given comparison, say differences between the strains at a given time point, we have to do another test strictly comparing those:

```
#just look at one gene to save space :)
results(ddsTC, name="strainmut.minute30", test="Wald")["SPBC2F12.09c",]
log2 fold change (MLE): strainmut.minute30
Wald test p-value: strainmut.minute30
DataFrame with 1 row and 6 columns
baseMean log2FoldChange lfcSE
<numeric> <numeric> <numeric>
SPBC2F12.09c 174.671161802578 -2.60046902875453 0.634342916342924
stat pvalue padj
<numeric> <numeric> <numeric>
SPBC2F12.09c -4.09946885471126 4.14099389594956e-05 0.27993118736619
```

- This compares the mutant strain to the wildtype strain at minute 30 (i think?). So the LFC represented here are the mutant expression values compared to the wildtype. This will generate a completely different p-value/padj than the previous LRT because its a completely different test comparing different things?

IF, that's right, then let's keep going down this rabbit hole:

What do the LFC in the heat map represent?

For example, if we look at SPBC2F12.09c, in the first column of the heatmap (minute

*15*vs_0) what samples are being compared? Is this all time 15 samples (mutant and wildtype) compared to all time 0 samples (mutant and wildtype)? That is, this gene was upregulated at 15 minutes compared to time 0. This would be generated by running:

```
res15 <- results(ddsTC, name="minute_15_vs_0", test="Wald")
res15["SPBC2F12.09c",]
log2 fold change (MLE): minute 15 vs 0
Wald test p-value: minute 15 vs 0
DataFrame with 1 row and 6 columns
baseMean log2FoldChange lfcSE
<numeric> <numeric> <numeric>
SPBC2F12.09c 174.671161802578 6.51883346980906 0.540737251817487
stat pvalue padj
<numeric> <numeric> <numeric>
SPBC2F12.09c 12.0554547479361 1.81527817129386e-33 2.56031468032063e-31
```

- Then, if we look at strainmut.minute15 for that same gene, we see that it is blue (down-regulated). Does this mean that it was down-regulated in the mutants at time 15 compared to the wildtype? If so, would the padj value for this comparison be calculated as so:

```
results(ddsTC, name="strainmut.minute15", test="Wald")["SPBC2F12.09c",]
log2 fold change (MLE): strainmut.minute15
Wald test p-value: strainmut.minute15
DataFrame with 1 row and 6 columns
baseMean log2FoldChange lfcSE
<numeric> <numeric> <numeric>
SPBC2F12.09c 174.671161802578 -2.57413709070726 0.64079390784322
stat pvalue padj
<numeric> <numeric> <numeric>
SPBC2F12.09c -4.01710606046689 5.89172116301028e-05 0.060423288283765
```

- This padj value is greater than 0.05, so with a FDR cutoff value of 0.05, would you consider this LFC significant? What if the LFC presented in the heat map do not have significant padj values from their respective pair-wise Wald tests? Can we still 'believe' the LFC?

I apologize for such a trivial question, but I would like to really fully understand this :) Thanks for your time!

Courtney

Heatmap of log2 fold changes for genes with smallest adjusted p value. The bottom set of genes show strong induction of expression for the baseline samples in minutes 15-60 (red boxes in the bottom left corner), but then have slight differences for the mutant strain (shown in the boxes in the bottom right corner).

**23k**• written 6 weeks ago by courtney.stairs •

**0**