DIA-NN Data Vignette

[abadgerw]

@yafeng Thank you for a great tool. I just wanted to check in on the status of the vignette you mentioned in a previous thread using your differential expression method on an output from DIANN? I am just a bit unclear as to what information I should be using as input from the reports.

[abadgerw]

@yafeng I just wanted to circle back and see if you had some insight into this query as I'd love to be able to use your fantastic tool?

[yafeng]

@abadgerw Hi, I missed your last comments. Sorry for that.

To use the DIA-NN output for DEqMS analysis, I have some tips for you.

Use the main report (report.tsv), the relevant columns are Precursor.Id, Genes (or Protein.Ids), Proteotypic, Precursor.Normalised.

If the data is not filtered based on the Q-value, use the Q.Value or Lib.Q.Value column (for library-free and spectra-library modes, respectively) for filtering.
The output is in the long format and should be transformed into a wide format.

Next, use MedianSweeping or MedianSummary function in DEqMS to group precursors into protein level quantity.
After you get a protein table, and a precursor count table for all proteins , you should be able to proceed to differential analysis.

if you get trouble, send me a short version of your DIA-NN output, I will write the R code for you.

Yafeng

[abadgerw]

Thanks, @yafeng! Is there a reason to roll up peptides to proteins using median sweeping rather than using the default method in DIA-NN that generates protein-level data and then extracting the precursor counts from the report.tsv file? I believe the new version of DIA-NN has a newly optimized method for quantification which was used for my dataset.

[yafeng]

@abadgerw，no, it is not necessary to use median sweep, the default DIA-NN protein level quant can be used directly . you can use the report.gg_matrix.tsv (gene-level quant) or report.pg_matrix.tsv(protein level quant), the relevant columns are Genes (or Protein.Group and Protein.Ids), MaxLFQ, precursor count.
The analysis workflow is similar to use MaxLFQ protein level quant as input.

Yafeng