DESeq2 with proteomics data
2
1
Entering edit mode
shyu ▴ 10
@shyu-16675
Last seen 4.7 years ago

Dear all,

Recently, I am using DESeq2 for differential analysis of proteomics data. But I got an error:

invalid class “DESeqDataSet” object: NA values are not allowed in the count matrix

I checked my input table, it contains no "NA".

Afterwards, I found that somebody had a similar error as mine ( DESeq2 Error NA values ). Based on the response in this post, I also found out my input table contains some values which are higher than 10 billion. Thus, I do a normalization to reduce the number of values. Then, the error is gone, and I got the results.

I would like to confirm that does DESeq2 not accept the values which is extremely high? But this kind of high value is quite common in the intensity of proteomics.

Thank you,

Silas

deseq2 • 3.3k views
3
Entering edit mode
@mikelove
Last seen 1 day ago
United States

We do not have support in DESeq2 for proteomics data.

There have been threads on this before. It's not simple count data as I understand (depending on actually what assay is performed...).

1
Entering edit mode
sven.schenk ▴ 20
@svenschenk-15837
Last seen 3.9 years ago
Vienna

We also tried using DESeq on proteomics data, and this gave extrmely high numbers of rejections (of the null), which didnt seem to be realistic.

The difference between RNASeq and proteomics data is that protein intensities are continuous data and not simple counts as are RNASeq data. This, I guess poses a problem for DESeq and the normalisation algorithm.

For proteomics data Id recomment to use either LIMMA or the ROTS package, both of which perform similar and, importantly, can deal with missing data points.