DESeq2 factors and factor level Input

[bveyb]

Dear Galaxy Support Team,

if I’m running DESeq2 with a merged data collection containing

3 replicates (column) with group tags group:WT group:control,

3 replicates (column) with group tags group:hipp5 group:control,

3 replicates (column) with group tags group:WT group:heat and

3 replicates (column) with group tags group:hipp5 group:heat

and choosing

‘condition’ as Factor 1 with the

1 factor level > group:hipp5

2 factor level > group:WT and

‘treatment’ as Factor 2 with the

1 factor level > group:heat

2 factor level > group:control

I obtain only analysis for the first Factor.

What is the reason for? Do I have to select a table with all factor data within the menue

“(Optional) provide a tabular file with additional batch factors to include in the model.”,

because there is “Nothing selected” at the moment.

Further, DESeq2 does run into error, if I’m choosing

‘condition’ as Factor 1 with

1 factor level > group:hipp5 and group:control

2 factor level > group:WT and group:control

3 factor level > group:hipp5 and group:heat

4 factor level > group:WT and group:heat

So it seems to me, choosing two group tags for one factor level isn’t possible.

Thank you for your help,

regards, Olaf.

P.S.: Could it be a bug within DESeq2 implementation? Because the following Input:

does produce the following results:

1. Factor in Input was “Condition” and 2. Factor in Input was “Treatment”,

but in results ‘other factors in design: Condition’ Why that???

[jennaj]

Welcome @bveyb

Hopefully we can help!

@bveyb

I obtain only analysis for the first Factor.

Yes, DESeq2 only reports results for one comparison at a time. If you want to explore the other combinations, you can rerun the tool again by rearranging the samples.

@jennaj

The first factor you input is considered as the primary factor that affects gene expressions. Optionally, you can input one or more secondary factors that might influence your experiment. But the final output will be changes in genes due to primary factor in presence of secondary factors.

You may also want to expore Limma and the use of contrasts. Warning that this has assumptions about the samples, so be sure to learn about how these tools differ! The vignettes, but also try their support forum for discuss from the authors like this: https://support.bioconductor.org/post/search/?query=deseq2+limma

The authors know about Galaxy and helped us to wrap the tools! But if you have Galaxy-specific questions, please come back here with new questions.

@bveyb

So it seems to me, choosing two group tags for one factor level isn’t possible.

Correct, the Galaxy tool form will only allow one group tag to be applied per input area. I suppose you could create another group tag and combine the samples the way you would prefer? The output is always with respect to the primary factor in the presence of the other factors, and there aren’t breakouts for each sub-group, so merging sub-groups is something you could explore!

@bveyb

but in results ‘other factors in design: Condition’ Why that???

To see the full design table, click into your Job Information Details page for one of the outputs.

1. In the inputs/parameters section, there will be a table with the information from the tool form summarized. This is a good way to keep track of which input was entered where on the form, all in a condensed table view.

2. In the job logs section (click to expand each), you’ll find the processing outputs from the original Bioconductor tool. Comparing the design table with the input summary table will help you to understand from where the input data is being sorted into the table. This connects data from the tool form through to DESeq2 itself.

Please let us know if this helps! Any questions, you are welcome to share back those two views together for context (screenshots or shared history) and we can try to help more!

---

I’ll also link in some resources…

Production workflows can be found at IWC Workflow Library! https://iwc.galaxyproject.org/. There is a base pair you might be interested in: reads to counts, then another for differential expression and visualization. Transcriptomics / Workflow List

Then, more as a reference for anyone reading later, since you are doing most of this already. → Hands-on: Group tags for complex experimental designs / Group tags for complex experimental designs / Using Galaxy and Managing your Data (#using-group-tags-in-tool-eg-deseq2)

And, see the Reference Based (DESeq2) and End-to-End (Limma) tutorials for more examples and discussion.- > GTN Transcriptomics / Tutorial List

[bveyb]

@jennaj

To see the full design table, click into your Job Information Details page for one of the outputs.

1. In the inputs/parameters section, there will be a table with the information from the tool form summarized. This is a good way to keep track of which input was entered where on the form, all in a condensed table view.
2. In the job logs section (click to expand each), you’ll find the processing outputs from the original Bioconductor tool. Comparing the design table with the input summary table will help you to understand from where the input data is being sorted into the table. This connects data from the tool form through to DESeq2 itself.

Dear jennaj, thank you for your help.

I took your suggestion and had a look in the ‘dataset details’ and

I’m still wondering, why the ‘tool standard output’ does mention ‘Condition’

instead of ‘Treatment’ for ‘other factors in design’ ?

With kind regards,

Olaf.

[jennaj]

More feedback next week!

[bveyb]

Dear jennaj,

thank you for your last reply and the examples for sample sheet input, the design formula.

Now I understand, that my multi factor input is correct and the DESeq2 Job output has some misleading information (output for ‘other factors in design’ → is same like the primary factor). Nevertheless the ‘design formula’ shows me that the second factor is used for multi factor analysis.

All the best,

Olaf.