directionality of the comparison from Deseq2 output

[shalinisingh65]

Hello,

Please can I ask about the directionality of the comparison from Deseq2 output. The following text is there but not very clear to me, X_vs_Y does not corelate to condition B and A:

"For each comparison (specified with --deg or, per default, all possible pairwise comparisons in one direction), a new folder X_vs_Y is created. This also describes the direction of the comparison, e.g. the log2FoldChange describes the change of a gene under condition B with respect to the gene under condition A."

[hoelzer]

Hi! Oh, good eye. I think this should be

"For each comparison (specified with --deg or, per default, all possible pairwise comparisons in one direction), a new folder X_vs_Y is created. This also describes the direction of the comparison, e.g., the log2FoldChange describes the change of a gene under condition Y with respect to the gene under condition X. For example, a log2FoldChange of -2 for gene A would tell you that this gene is 2-fold downregulated when we compare condition X vs. condition Y. The gene A is lower expressed in condition Y."

@MarieLataretu @fischer-hub can you confirm that? Because we switched the comparison at some point, I'm not 100% sure. When clarified, we can update the README accordingly to make this clear. Thanks!

[fischer-hub]

I think this is when we changed this? But from the description in the PR I would understand that a log2FoldChange of -2 for gene A would tell you that this gene is 2-fold downregulated with condition X. The gene A is lower expressed in condition X?

If logFC < 0, then **down** in infected and   **up** in mock

for infected_vs_mock

But not sure..

[hoelzer]

@fischer-hub yes! I think you're actually right. But still not 100% w/o checking... I just looked into some recent data I calculated.

Pairwise comparison between CBS14_P2_vs_SBY92_P1 (that's the folder name as well, so X = CBS14_P2 and Y = SBY92_P1 in this example)

Here, a gene A was found as highly differential expressed w/ a log2 FC of 10.14.

Now, the question is: is the gene high expressed in X = CBS14_P2 or Y = SBY92_P1?

The image shows the triplicates belonging to the two conditions X and Y. We can see, that the gene A is highly expressed in X = CBS14_P2 but not at all in Y = SBY92_P1.

So based on that, the README text should be:

"For each comparison (specified with --deg or, per default, all possible pairwise comparisons in one direction), a new folder X_vs_Y is created. This also describes the direction of the comparison, e.g., the log2FoldChange describes the change of a gene A under condition Y with respect to the gene under condition X. For example, a log2FoldChange of +2 for gene A would tell you that this gene is 2-fold upregulated when we compare condition X vs. condition Y. The gene A is higher expressed in samples belonging to condition X."

Agree? If so, we can update the README accordingly.

[fischer-hub]

@hoelzer yes, seems right! I'll update the README :)

[hoelzer]

Perfect, thx!