Help with the DIA-NN settings and the DBIA modification

[SNgalaxy]

Hello there,

I am running a test on DIA-NN and I have issues getting the modifications on the peptide for the DBIA. My data is from a DIA public database and the code I am running is as below:

I generated the specLib first.
Then I uploaded the file, the speclIb and the fasta files.

1. My very first question is what should be precursor ion generation mode in the GUI:
   Library search off OR
   Prediction from library?

2. In the output section of the GUI, should I enable the Generate spectral library again?

3. Is it the correct way of defining DBIA modification? I searched everywhere on the manual and I could not find the correct way. Would you please clarify it?

Thank you for your help

diann.exe --f "D:\PXD054127\DC2461_ELB00032_20220318_DIA_H00AGKI_P02_B02_S2-B2_1_2005.d" --lib "D:\PXD054127\fasta_speclib\report-lib.predicted.speclib" --threads 6 --verbose 3 --out "C:\Users\Administrator\Desktop\DIANN_result_test\run5\report.parquet" --qvalue 0.01 --matrices --out-lib "C:\Users\Administrator\Desktop\DIANN_result_test\test-lib.parquet" --gen-spec-lib --fasta "C:\Users\Administrator\Desktop\fragpipe\2025-10-30-decoys-reviewed-contam-UP000005640_2.fasta" --met-excision --min-pep-len 7 --max-pep-len 30 --min-pr-mz 300 --max-pr-mz 1800 --min-pr-charge 1 --max-pr-charge 4 --min-fr-mz 200 --max-fr-mz 1800 --cut K*,R* --missed-cleavages 1 --mass-acc 15 --mass-acc-ms1 10 --reanalyse --rt-profiling **--var-mods 2 --var-mod UniMod:4,57.02146,C --var-mod DBIA:DBIA,296.1818,C**

DIA-NN 2.3.0 Academia (Data-Independent Acquisition by Neural Networks)
Compiled on Sep 27 2025 22:03:55
Current date and time: Tue Dec 16 10:15:53 2025
CPU: GenuineIntel Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz
SIMD instructions: AVX AVX2 FMA SSE4.1 SSE4.2
Logical CPU cores: 12
Thread number set to 6
Output will be filtered at 0.01 FDR
Precursor/protein x samples expression level matrices will be saved along with the main report
A spectral library will be generated
N-terminal methionine excision enabled
Min peptide length set to 7
Max peptide length set to 30
Min precursor m/z set to 300
Max precursor m/z set to 1800
Min precursor charge set to 1
Max precursor charge set to 4
Min fragment m/z set to 200
Max fragment m/z set to 1800
In silico digest will involve cuts at K*,R*
Maximum number of missed cleavages set to 1
MBR enabled; .quant files will only be saved to disk during the first pass
The spectral library (if generated) will retain the original spectra but will include empirically-aligned RTs
Maximum number of variable modifications set to 2
Modification UniMod:4 with mass delta 57.0215 at C will be considered as variable
Modification DBIA:DBIA with mass delta 296.182 at C will be considered as variable
Cannot find a UniMod modification match for DBIA:DBIA: 140.081 minimal mass discrepancy; using the original modificaiton name
Mass accuracy will be fixed to 1.5e-05 (MS2) and 1e-05 (MS1)
WARNING: peptidoform scoring enabled because variable modifications have been declared; to disable, use --no-peptidoforms
The following variable modifications will be localised: UniMod:4 DBIA:DBIA
WARNING: MBR turned off, at least two files are required

1 files will be processed
[0:00] Loading spectral library D:\PXD054127\fasta_speclib\report-lib.predicted.speclib
[0:07] Library annotated with sequence database(s): C:\Users\Administrator\Desktop\fragpipe\2025-10-30-decoys-reviewed-contam-UP000005640_2.fasta
[0:08] Spectral library loaded: 40863 protein isoforms, 59253 protein groups and 8607896 precursors in 2679296 elution groups.
[0:08] Loading protein annotations from FASTA C:\Users\Administrator\Desktop\fragpipe\2025-10-30-decoys-reviewed-contam-UP000005640_2.fasta
[0:09] Annotating library proteins with information from the FASTA database
[0:09] Gene names missing for some isoforms
[0:09] Library contains 25451 proteins, and 25112 genes
[0:16] Initialising library
[0:35] Library RAM: 919MB (core), 1163MB (fragments), 161MB (names), 36MB (protein_ids), 24MB (protein

[vdemichev]

Hi,

1. If you already have a library and don't want to change anything in there, then it's Library search / Off.
2. Normally, yes, creation of empirical library is part of MBR.
3. Any modification can be defined with --var-mod, syntax is incorrect, but it seems DIA-NN reads it as it should in this case anyway.

--rt-profiling **--var-mods 2

Why double asterisk here and at the end?

These two sections may be helpful:
https://github.com/vdemichev/DiaNN?tab=readme-ov-file#getting-started-with-dia-nn
https://github.com/vdemichev/DiaNN?tab=readme-ov-file#fine-tuning-prediction-models

Best,
Vadim

[vdemichev]

With DBIA should be fairly straightforward:

• Generate a predicted lib with --var-mod DBIA,296.1818,C.
• Analyse one or several runs with this library, with Speed: RT/IM filtering set to Relaxed (slow); Generate spectral library box should be on. During analysis also supply --var-mod DBIA,296.1818,C.
• Fine-tune predictors as described by the link above, using this empirical lib.
• Re-generate the predicted lib using fine-tuned predictors (referencing them with --tokens, --rt-model, --im-model and --fr-model).
• Analyse the entire experiment with the new predicted lib, make sure MBR is enabled.

Hope this helps!

Best,
Vadim

[SNgalaxy]

Thank you so much for the fast response.

Generate a predicted lib with --var-mod DBIA,296.1818,C.

In the above section, if I want to consider UniMod:4 as variable too, then should I add it as an extra comment to the additional options box?

something like below:
--var-mod UniMod:4 --var-mod DBIA,296.1818,C.

Thanks

[vdemichev]

In the above section, if I want to consider UniMod:4 as variable too, then should I add it as an extra comment to the additional options box?

Uncheck the respective box in the GUI and add --var-mod UniMod:4,57.021464,C.

[SNgalaxy]

Thank you !

[vdemichev]

Btw, forgot to mention, if you use --tune-fr, please do generate the empirical lib that you use for tuning with library generation set to Full profiling.

Best,
Vadim

[SNgalaxy]

Hello Vadim,

Thank you for your help in setting up the process. I just have a final question for you. The steps that I took are as below:

1. generate the predicted library using --var-mod --var-mod DBIA,296.1818,C.
2. generate the empirical library using three runs:

diann.exe --f D:\PXD054127\DC2461_ELB00032_20220318_DIA_H00AGKI_P02_B02_S2-B2_1_2005.d --f D:\PXD054127\DC2461_ELB00032_20220318_DIA_H00AGKI_P02_C12_S2-C12_1_2033.d --f D:\PXD054127\DC2461_ELB00032_20220318_DIA_H00AGKJ_P03_B12_S3-B12_1_2228.d --lib D:\PXD054127\fasta_speclib\report-lib.predicted.speclib --threads 10 --verbose 1 --out C:\Users\Administrator\Desktop\DIANN_result_test\emperical_library_generation1\report.parquet --qvalue 0.01 --matrices --no-rt-window --out-lib C:\Users\Administrator\Desktop\DIANN_result_test\emperical_library_generation1\report-lib.parquet --gen-spec-lib --fasta camprotR_240512_cRAP_20190401_full_tags.fasta --cont-quant-exclude cRAP- --fasta C:\Users\Administrator\Desktop\fragpipe\2025-10-30-decoys-reviewed-contam-UP000005640_2.fasta --pre-search --pre-filter --met-excision --min-pep-len 7 --max-pep-len 30 --min-pr-mz 300 --max-pr-mz 1800 --min-pr-charge 1 --max-pr-charge 4 --min-fr-mz 200 --max-fr-mz 1800 --cut K*,R* --missed-cleavages 1 --mass-acc 15 --mass-acc-ms1 10 --reanalyse --var-mods 2 --var-mod DBIA,296.1818,C --var-mod UniMod:4,57.021464,C

The GUI setting in the precursor icon generation was as below:
enabling contaminants
Mode : InfinitDIA pre-search
Library generation: Full profiling
speed: Relaxed
Is it mandatory to enable the proteoforms scoring in this step on the GUI?
Is it okay to enable the MBR in this step?

3. Fine tuning:

diann.exe --lib --threads 6 --verbose 1 --out C:\Users\Administrator\Desktop\DIANN_result_test\tuning\report.parquet --qvalue 0.01 --matrices --out-lib C:\Users\Administrator\Desktop\DIANN_result_test\tuning\report-lib.parquet --gen-spec-lib --fasta camprotR_240512_cRAP_20190401_full_tags.fasta --cont-quant-exclude cRAP- --fasta C:\Users\Administrator\Desktop\fragpipe\2025-10-30-decoys-reviewed-contam-UP000005640_2.fasta --pre-search --pre-filter --met-excision --min-pep-len 7 --max-pep-len 30 --min-pr-mz 300 --max-pr-mz 1800 --min-pr-charge 1 --max-pr-charge 4 --min-fr-mz 200 --max-fr-mz 1800 --cut K*,R* --missed-cleavages 1 --peptidoforms --reanalyse --rt-profiling --tokens C:\Users\Administrator\Desktop\DIANN_result_test\emperical_library_generation1\report-lib.dict.txt --rt-model C:\Users\Administrator\Desktop\DIANN_result_test\emperical_library_generation1\[report-lib.tuned_rt.pt](http://report-lib.tuned_rt.pt/) --im-model C:\Users\Administrator\Desktop\DIANN_result_test\emperical_library_generation1\[report-lib.tuned_im.pt](http://report-lib.tuned_im.pt/) --mod DBIA,296.1818,C --var-mods 2 --var-mod DBIA,296.1818,C --var-mod UniMod:4,57.021464,C

4. Re-generate the predicted lib using fine-tuned predictors

diann.exe --lib "" --threads 6 --verbose 1 --out "C:\Users\Administrator\Desktop\DIANN_result_test\calibration\report.parquet" --qvalue 0.01 --matrices --out-lib "C:\Users\Administrator\Desktop\DIANN_result_test\calibration\tuned.parquet" --gen-spec-lib --predictor --fasta camprotR_240512_cRAP_20190401_full_tags.fasta --cont-quant-exclude cRAP- --fasta "C:\Users\Administrator\Desktop\fragpipe\2025-10-30-decoys-reviewed-contam-UP000005640_2.fasta" --fasta-search --met-excision --min-pep-len 7 --max-pep-len 30 --min-pr-mz 300 --max-pr-mz 1800 --min-pr-charge 1 --max-pr-charge 4 --min-fr-mz 200 --max-fr-mz 1800 --cut K*,R* --missed-cleavages 1 --rt-profiling --tokens C:\Users\Administrator\Desktop\DIANN_result_test\emperical_library_generation1\report-lib.dict.txt --rt-model C:\Users\Administrator\Desktop\DIANN_result_test\emperical_library_generation1\[report-lib.tuned_rt.pt](http://report-lib.tuned_rt.pt/) --im-model C:\Users\Administrator\Desktop\DIANN_result_test\emperical_library_generation1\[report-lib.tuned_im.pt](http://report-lib.tuned_im.pt/) --var-mods 2 --var-mod DBIA,296.1818,C --var-mod UniMod:4,57.021464,C

In the GUI application, the Generate spectral library is on

Thank you for your help again

[vdemichev]

Is it okay to enable the MBR in this step?

It's essential for InfinDIA, yes.

[SNgalaxy]

Hi Vdemichev,

Than you for you help so far. I have reabalyzed the whole data from the paper, with the spec-lib generated after tuning. Though I do not get the sit.report file and I would like to see the modification site. I also do not see the position of the sites in the report.paraquet file. Should I have run the analysis uploading the fasta file too? I did not do so after the above final three steps. Appreciate your help

[vdemichev]

Hi,

Yes, you need to specify a FASTA to get a site_report, i.e. DIA-NN needs to know the protein sequences to match peptides to protein sites. But localisation in general is only be relevant for peptides with multiple cysteines.

Best,
Vadim