Some questions about DIA-NN output

[wumingxiaocuier]

Hello, I used DIA-NN's library free and deep learning mode to identify Astral DDA data and have the following questions about the results:

1. Some PSMs do not have corresponding peaks of the identified peptide's m/z in MS1 (given by the instrument of MS2)
2. The fragmentation center of the precursor provided by MS2 differs significantly from the m/z of the peptide identified on this MS2 I am not quite sure about the identification strategy of DIA-NN in library free, deep learning mode, which makes me uncertain whether the identified peptide in DIA-NN really exist, or these two issues are bugs?

[wumingxiaocuier]

I read the paper of DIA-NN (DIA-NN: neural networks and interference correction ...), which made me more confused. ~>_<~ What's the meaning of "Chromatograms are extracted for each target and decoy precursor and the respective fragment ions. Potential elution peaks are identified and ...". Is the peptides in silico are matched to the elution peaks in MS1 (even they may not have the corresponding MS2)? But how to explain the match of their fragment ions...... ×﹏×

[vdemichev]

Hi, DIA-NN is not meant to be used for DDA data. While it can work with some DDA data, this is purely by chance basically :)

Best, Vadim

[wumingxiaocuier]

Hello, “the potential elution peaks are identified”, what's the basis of the peak to be regarded as the precursor's potential elution peak?

[vdemichev]

Primarily based on co-detection of multiple fragment ions at the same RT.