DIA-NN - a fast and easy to use tool for processing data-independent acquisition (DIA) proteomics data.
DIA-NN implements deep neural networks to improve precursor ion identification.
DIA-NN now also supports library-free search and spectral library generation.
DIA-NN: neural networks and interference correction
enable deep proteome coverage in high throughput
Vadim Demichev, Christoph B. Messner, Spyros I. Vernardis, Kathryn S. Lilley, Markus Ralser
https://www.nature.com/articles/s41592-019-0638-x
DIA-NN encompasses all stages of DIA-MS data processing in a single program.
As input it takes raw data files and a spectral library / FASTA database.
A report with protein and precursor ion quantities is produced.
DIA-NN manual: https://github.com/vdemichev/DiaNN/blob/master/DIA-NN%20GUI%20manual.pdf
Please also check the commands listed below.
-
Deep learning-based generation of high quality spectral libraries in silico (Windows only). This boosts library-free search efficiency.
-
Analysis logs (which also specify the DIA-NN version used and the search settings) are now saved automatically.
Download and run the automatic installer: DIA-NN-Setup.msi.
For .wiff support, first download and install ProteoWizard (http://proteowizard.sourceforge.net/download.html - choose the version (64-bit) that supports "vendor files"), then use the installer (DIA-NN-Setup.msi) and specify the ProteoWizard directory (e.g. C:\Program Files (x86)\ProteoWizard\ProteoWizard [version], where [version] is the ProteoWizard version number, e.g. 3.0.11537) as the installation directory.
Raw data files: Sciex .wiff, Thermo .raw, .mzML or .dia (format used by DIA-NN to store spectra).
Reading Thermo .raw files requires Thermo MS File Reader (https://thermo.flexnetoperations.com/control/thmo/login?nextURL=%2Fcontrol%2Fthmo%2Fdownload%3Felement%3D6306677) to be installed. It is essential to use specifically the version by the link above (3.0 SP3).
The .mzML files should be centroided and contain data as spectra (e.g. SWATH) and not chromatograms.
Spectral library: comma-separated (.csv) or tab-separated (.tsv, .xls or .txt) or .speclib (compact format used by DIA-NN) file.
diann.exe [commands]
Commands can be supplied in arbitrary order.
--f <data file>
Specifies a data file to be processed. Use --f for each file to be processed.
--lib <spectral library file>
Specifies the spectral library. Example:
diann.exe --f run1.mzML --f run2.mzML --lib yeast.tsv
--cfg <config file>
Specifies a file with a set of commands.
--dir <folder>
Specifies a folder containing raw files to be processed. All files in the folder must be in .raw, .mzML or .dia format.
--threads <thread number>
Specifies the number of CPU threads to be used.
--convert
With this option DIA-NN converts .mzML files (specified using the --f command) to the .dia format. Unlike .mzML files, .dia files can be loaded quickly (seconds), so it is recommended to convert files that are going to be analysed multiple times.
--ext <string>
Add a string to the end of each file name (specified with --f).
--prefix <string>
Add a string to the beginning of each file name.
--out <output file>
Specifies the output file; by default, the output is saved to quant.tsv in the current working directory.
diann.exe --threads 4 --f run1 --f run2 --lib yeast.tsv --prefix C:\Data\ --ext .mzML --out run1_2.tsv
This following command is supported in development versions uploaded on 20/12/2019 or later.
--vis <N>,<Peptide 1>,<Peptide 2>,...
Specifies a number of peptides for which chromatograms of length >= N will be extracted and saved as a text table. These allow for PSM visualisation using third-party software (e.g. R or Python).
diann.exe --f run1.mzML --f run2.mzML --lib yeast.tsv --out report.tsv --vis 20,KVYPDVLYTSK,TAIEGSYIDK,DSATHELTK
For a full list of supported commands see the arguments() function in /src/diann.cpp.
Windows: A Visual C++ solution file is provided with the source code. Changing the SDK to a Windows 10 one in the project settings might be required. Tested on Windows 7 and 10. Configuration should be set to "Release" (source code for deep learning-based generation of spectral libraries (PyTorch) is not included, so this functionality is only supported in the official DIA-NN builds).
Linux (GCC 7 or later required):
git clone https://www.github.com/vdemichev/diann
cd diann/mstoolkit
make
Bash scripts for building without .mzML support (for this, uncomment "//#undef MSTOOLKIT" after "#ifdef LINUX") are also provided in the root directory.
This is a simple tutorial which covers the generation of a spectral library from gas-phase fractionated DIA data and its use to analyse other DIA runs (using DIA-NN 1.7.5).
-
Download all the files from https://osf.io/w5dr6/files/?view_only=00c8a68bfb824835b7fa304e31922ffa - these are 9 yeast runs on TripleTOF 6600 converted to .dia format: select "OSF Storage (Germany - Frankfurt)" and then choose "Download as zip" above (7.6 Gb total). Unpack the zip file.
-
Download the UniProt canonical S.cerevisiae proteome from https://www.uniprot.org/proteomes/UP000002311 (Under "Components" click "Download" and then "Go").
-
Download the collection of peptides known to be detectable in yeast samples from http://www.peptideatlas.org/builds/: find the Yeast ("Build name" column) Mar 2013 ("Date" column) entry, right click on "APD_Sc_all.fasta" in the "Peptide Sequences" column and choose "Save link as".
-
Launch DIA-NN. Click Add raw data in the Input panel, select the 8 gas-phase fractionation runs (with names 400-500, 495-600, ... 1095-1250). Click Add FASTA and select APD_Sc_all.fasta. Check FASTA digest for library-free search / library generation in the Precursor ion generation panel. Make sure Deep learning-based spectra and RTs prediction (same panel) as well as Generate spectral library (Output panel) are checked. Choose where to save the Main output file (e.g. report.tsv in the same folder). Choose where to save the Output library (e.g. lib.tsv in the same folder). Click Add to pipeline (just above the pipeline window).
-
Copy-paste the Output library field to the Spectral library field in the Input panel. Uncheck FASTA digest for library-free search / library generation, Deep learning-based spectra and RTs prediction and Generate spectral library. Click Clear list next to the Add raw data button and then add pre_qc_1.wiff.dia (this is 23-min microflow gradient SWATH run on Sciex TripleTOF 6600). Click Clear list below the Add FASTA button, click Add FASTA and select the UniProt sequence database, check Reannotate (to add protein information from the sequence database to the spectral library). Click Add to pipeline. Click Execute (below the pipeline window). On Intel i3-8350K, the whole pipeline (creation of the spectral library plus its use to analyse the 23-minute run) takes about 17 minutes and results in ~25000 precursors identified at 1% FDR.
Comments:
-
DIA-NN can create libraries from any DIA/SWATH runs (not necessarily gas-phase fractionation runs). Sometimes even a single long-gradient run might be good enough.
-
Library-free search was used to create a spectral library. Regular spectral library-based search can also be used to generate new libraries. This is useful e.g. for refining spectra and retention times in a public library.
-
The peptide list (APD_Sc_all) was used primarily to speed up the analysis (as this is a tutorial), it is OK to use the full sequence database instead.
-
Each of the pipeline steps can be executed separately (select a step and click Run). Pipeline creation is also optional (can just click Run to execute the current configuration).
Many thanks to Christoph Messner, Spyros Vernardis, Kathryn Lilley and Markus Ralser as well as to the members of the Ralser lab (Francis Crick Institute and Charité - Universitätsmedizin Berlin) for their continuous support in improving DIA-NN and expanding its functionality.
Prosit https://www.nature.com/articles/s41592-019-0426-7 online service (part of ProteomeTools) https://www.proteomicsdb.org/prosit/ has been utilised when developing DIA-NN versions uploaded after 24/01/2020.