Question on DF calculation for GroupComparison with technical replicates

[Martin Gordon]

Hi everyone, 

Thanks for developing this great tool. I have a question on how MSstats uses technical replicates in 

My experimental setup is as follows:

Two conditions (mutant, WT) with two and three subjects respectively. Each subject has 3 technical replicates for 15 MS runs in total.

For differential testing, I treated this as a simple group comparison. Different subjects in each condition are distinguished in BioReplicate column (eg. WT.1,  Mut.1, WT.2, Mut.2 etc...). 

When looking at the output, I was surprised to see the degrees of freedom were higher than expected (histogram attached). I'd like to have more of an understanding of how DF is calculated in this case; are the technical replicates treated as independent observations?

Thank you,

Martin

[Mateusz Staniak]

Hi,

number of degrees of freedom is the same as [residual DF] in a model with Condition and BioReplicate terms.

Kind regards,

Mateusz

[Frank Fabela]

Hello,

To build off of Martin's question, why do the DF values vary between proteins if the number of BioReps and Conditions are constant?

Thank you,

Frank

[Mateusz Staniak]

Hi,

I would need to see and example, but missigness could play a role here.

Kind regards,

Mateusz

[Frank Fabela]

Hello,

After some digging in the fitted model output, I understood what you meant in your previous answer. Thank you!

Is the residual DF used only for unbalanced designs? The MSstats 4.0 paper gives examples on balanced designs and shows the DF for group comparison as: #Conditons * (#BioReps-1).

-Frank