why unchecking "Deep learning based spectra" checkbox give more idetifications

[amrelguo]

I searched DIA endogenous peptide data against constructed DDA spectral library with and without checking "Deep learning-based spectra" check box.
the results showed that more identification can be obtained without ticing "Deep learning-based spectra" checkbox. i wondering why this happened and if these unique identificaions can be considered as significant ones or not.
thanks in advance.

[vdemichev]

Yes, very often empirical spectra, RTs and IMs can be better than predicted, this is expected.

[amrelguo]

Thanks so much for your quick response.

…

On Thu, May 18, 2023 at 6:15 PM Vadim Demichev ***@***.***> wrote: Yes, very often empirical spectra, RTs and IMs can be better than predicted, this is expected. — Reply to this email directly, view it on GitHub <#702 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A7546NFBXH7PXI46OUR3X3DXGXSCTANCNFSM6AAAAAAYF7YRQY> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[amrelguo]

despite I didn't tick "deep learning based spectra" checkbox, the deep learning was done with less number of targets and decoys than those used when i ticked. what is the reason?

report.log_NO_deep_learning.txt
report.log_with_deep_learning.txt