Analysis of diaPASEF Data
For this tutorial, we are using publicly available data from the publication: *Meier et al., Nature Methods (2020)*.
We will perform the following steps in this tutorial:
Spectral library generation using
Generating pseudo-spectra using diaTracer
Search pseudo-spectra using MSFragger
Validate PSMs using PeptideProphet and iProphet
Generate Spectral Library using EasyPQP
Generate Run-Specific iRT libraries
Targeted Data Extraction using OpenSwathWorkflow
Statistical Scoring and FDR Control using PyProphet
Data
The dataset is deposited in PRIDE under accession PXD017703. Specifically, we will work with the HeLa_Evosep_diaPASEF_RAW.zip experiment, focusing on the 200SPD_py8 experiment.
You should see the following files when you download and unzip the data (assuming you are in the HeLa_Evosep_diaPASEF_RAW/200SPD_py8 directory):
20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740.d
20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A1_1_2737.d
20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A2_1_2738.d
Software requirements
To follow along, you will need the following software installed:
Java (≥ v1.8) for running diaTracer and MSFragger
diaTracer (≥ v1.3.3) for generating pseudo-spectra from diaPASEF data
msfragger (≥ v4.3) for proteomics database searching
TPP (≥ v7.2.0) for PSM validation using PeptideProphet and iProphet
OpenMS (≥ v3.4) for various data processing steps including spectral library generation and running OpenSwathWorkflow
Python (≥ 3.10) with the following packages:
easypqp (≥ v0.1.53) (for spectral library generation)
diapysef (≥ v1.0.10) (for Bruker tdf file conversion)
pyprophet (≥ v3.0.2) (for statistical scoring and FDR control)
Docker (optional) if you prefer running the tools in a containerized environment
Make sure these tools are available in your PATH before proceeding.
Spectral Library Generation
To analyze diaPASEF data with OpenSwath, a spectral library is required. This can be generated using any of the tools mentioned in the Spectral library generation section. For this tutorial, we will use a spectral library generated using pseudo-spectra generated from diaTracer and using MSFragger for spectra database searching.
Note
If you prefer a GUI, you can also use FragPipe to generate the spectral library. FragPipe provides an integrated environment for running diaTracer, MSFragger, PeptideProphet/Percolator, iProphet, and EasyPQP. See the fragpipe section.
Note
In this tutorial, we are only using diaPASEF data. However, if you have ddaPASEF data as well, you can create a consensus library using both ddaPASEF and the pseudo-spectra from diaPASEF.
Generating pseudo-spectra using diaTracer
#!/bin/bash
mkdir -p library/
mkdir -p library/diatracer/
# Run diaTracer on all .d directories using all available CPU cores
find . -maxdepth 1 -name "*.d" -type d | parallel -j+0 \
'input_file={}; \
base_name=$(basename "$input_file"); \
log_file="library/diatracer/diatracer_${base_name}.log"; \
echo "Processing $base_name - Output going to $log_file"; \
java -jar /diaTracer.jar --dFilePath "$base_name" --workDir "library/diatracer/" --deltaApexIM 0.01 --deltaApexRT 3 --ms1MS2Corr 0.3 --massDefectFilter 0 --massDefectOffset 0.1 --RFMax 500 --threadNum 3 > "$log_file" 2>&1'
Note
The above command uses GNU Parallel to process all .d directories in the current directory, generating pseudo-spectra for each and saving the logs to individual files. To control the number of concurrent jobs, adjust the -j parameter in the parallel command, for example, -j4 to limit to 4 simultaneous jobs.
Search pseudo-spectra using MSFragger
You can retrieve a suitable fasta file from the UniProt website (Human reference proteome). Make sure to include common contaminants and reverse decoys in the fasta file. For the following tutorial, we use the following fasta file. If your fasta file does not contain decoys, you can generate and append a decoy database using OpenMS’s DecoyDatabase tool.
#!/bin/bash
mkdir -p library/msfragger/
wdir=$(pwd)
# Copy the fasta file to the msfragger directory since our param file points to the current relative path, which will be the msfragger directory when we run MSFragger
cp library_decoys.fasta library/msfragger/
cd library/msfragger/
# Run MSFragger on all pseudo-spectra mzML files using 1 job at a time
ls -1 $wdir/library/diatracer/*_diatracer.mzML | parallel -j1 \
'input_file={}; \
base_name=$(basename "$input_file" .mzML); \
log_file="library/msfragger/msfragger_${base_name}.log"; \
echo "Processing $base_name - Output going to $log_file"; \
java -Xmx16G -jar /MSFragger.jar $wdir/fragger_closed.params $wdir/"$base_name".mzML > "$log_file" 2>&1'
cd $wdir
fragger_closed.params
num_threads = 4 # Number of CPU threads to use.
database_name = library_decoys.fasta # Path to the protein database file in FASTA format.
precursor_mass_lower = -20 # Lower bound of the precursor mass window.
precursor_mass_upper = 20 # Upper bound of the precursor mass window.
precursor_mass_units = 1 # Precursor mass tolerance units (0 for Da, 1 for ppm).
data_type = 0 # Data type (0 for DDA, 1 for DIA, 2 for gas-phase fractionation DIA, 3 for DDA+).
precursor_true_tolerance = 20 # True precursor mass tolerance (window is +/- this value).
precursor_true_units = 1 # True precursor mass tolerance units (0 for Da, 1 for ppm).
fragment_mass_tolerance = 20 # Fragment mass tolerance (window is +/- this value).
fragment_mass_units = 1 # Fragment mass tolerance units (0 for Da, 1 for ppm).
calibrate_mass = 2 # Perform mass calibration (0 for OFF, 1 for ON, 2 for ON and find optimal parameters, 4 for ON and find the optimal fragment mass tolerance).
use_all_mods_in_first_search = 0 # Use all variable modifications in first search (0 for No, 1 for Yes).
decoy_prefix = DECOY_ # Prefix added to the decoy protein ID.
deisotope = 1 # Perform deisotoping or not (0=no, 1=yes and assume singleton peaks single charged, 2=yes and assume singleton peaks single or double charged).
deneutralloss = 1 # Perform deneutrallossing or not (0=no, 1=yes).
isotope_error = 0/1/2/3 # Also search for MS/MS events triggered on specified isotopic peaks.
mass_offsets = 0.0 # Creates multiple precursor tolerance windows with specified mass offsets.
mass_offsets_detailed = # Optional detailed mass offset list. Overrides mass_offsets if use_detailed_offsets = 1.
use_detailed_offsets = 0 # Whether to use the regular (0) or detailed (1) mass offset list.
precursor_mass_mode = selected # One of isolated/selected/corrected.
remove_precursor_peak = 1 # Remove precursor peaks from tandem mass spectra. 0 = not remove; 1 = remove the peak with precursor charge; 2 = remove the peaks with all charge states (only for DDA mode).
remove_precursor_range = -1.500000,1.500000 # m/z range in removing precursor peaks. Only for DDA mode. Unit: Th.
intensity_transform = 0 # Transform peaks intensities with sqrt root. 0 = not transform; 1 = transform using sqrt root.
activation_types = all # Filter to only search scans of provided activation type(s), separated by /. Allowed: All, HCD, CID, ETD, ECD.
analyzer_types = all # Filter to only include scans matching the provided analyzer type(s) in search, separated by /. Only support the mzML and raw format. Allowed types: all, FTMS, ITMS.
group_variable = 0 # Specify the variable used to decide the PSM group in the group FDR estimation. 0 = no group FDR; 1 = num_enzyme_termini; 2 = PE from protein header.
require_precursor = 1 # If required, PSMs with no precursor peaks will be discarded. For DIA data type only. 0 = no, 1 = yes.
reuse_dia_fragment_peaks = 0 # Allow the same peak matches to multiple peptides. For DIA data type only. 0 = no, 1 = yes.
write_calibrated_mzml = 1 # Write calibrated MS2 scan to a mzML file (0 for No, 1 for Yes).
write_uncalibrated_mzml = 0 # Write uncalibrated MS2 scan to a MGF file (0 for No, 1 for Yes). Only for .raw and .d formats.
write_mzbin_all = 0
mass_diff_to_variable_mod = 0 # Put mass diff as a variable modification. 0 for no; 1 for yes and remove delta mass; 2 for yes and keep delta mass.
localize_delta_mass = 0 # Include fragment ions mass-shifted by unknown modifications (recommended for open and mass offset searches) (0 for OFF, 1 for ON).
delta_mass_exclude_ranges = (-1.5,3.5) # Exclude mass range for shifted ions searching.
fragment_ion_series = b,y # Ion series used in search, specify any of a,b,c,x,y,z,Y,b-18,y-18 (comma separated).
ion_series_definitions = # User defined ion series. Example: "b* N -17.026548;b0 N -18.010565".
labile_search_mode = off # type of search (nglycan, labile, or off). Off means non-labile/typical search.
restrict_deltamass_to = all # Specify amino acids on which delta masses (mass offsets or search modifications) can occur. Allowed values are single letter codes (e.g. ACD) and '-', must be capitalized. Use 'all' to allow any amino acid.
diagnostic_intensity_filter = 0 # [nglycan/labile search_mode only]. Minimum relative intensity for SUM of all detected oxonium ions to achieve for spectrum to contain diagnostic fragment evidence. Calculated relative to spectrum base peak. 0 <= value.
Y_type_masses = # [nglycan/labile search_mode only]. Specify fragments of labile mods that are commonly retained on intact peptides (e.g. Y ions for glycans). Only used if 'Y' is included in fragment_ion_series.
diagnostic_fragments = # [nglycan/labile search_mode only]. Specify diagnostic fragments of labile mods that appear in the low m/z region. Only used if diagnostic_intensity_filter > 0.
remainder_fragment_masses = # [labile search_mode only] List of possible remainder fragment ions to consider. Remainder masses are partial modification masses left on b/y ions after fragmentation.
search_enzyme_name_1 = stricttrypsin # Name of the first enzyme.
search_enzyme_cut_1 = KR # First enzyme's cutting amino acid.
search_enzyme_nocut_1 = # First enzyme's protecting amino acid.
search_enzyme_sense_1 = C # First enzyme's cutting terminal.
allowed_missed_cleavage_1 = 2 # First enzyme's allowed number of missed cleavages per peptide. Maximum value is 5.
search_enzyme_name_2 = null # Name of the second enzyme.
search_enzyme_cut_2 = # Second enzyme's cutting amino acid.
search_enzyme_nocut_2 = # Second enzyme's protecting amino acid.
search_enzyme_sense_2 = C # Second enzyme's cutting terminal.
allowed_missed_cleavage_2 = 2 # Second enzyme's allowed number of missed cleavages per peptide. Maximum value is 5.
num_enzyme_termini = 2 # 0 for non-enzymatic, 1 for semi-enzymatic, and 2 for fully-enzymatic.
clip_nTerm_M = 1 # Specifies the trimming of a protein N-terminal methionine as a variable modification (0 or 1).
# maximum of 16 mods - amino acid codes, * for any amino acid, [ and ] specifies protein termini, n and c specifies peptide termini
variable_mod_01 = 15.9949 M 3
variable_mod_02 = 42.0106 [^ 1
# variable_mod_03 = 79.96633 STY 3
# variable_mod_04 = -17.0265 nQnC 1
# variable_mod_05 = -18.0106 nE 1
# variable_mod_06 = 4.025107 K 2
# variable_mod_07 = 6.020129 R 2
# variable_mod_08 = 8.014199 K 2
# variable_mod_09 = 10.008269 R 2
# variable_mod_10 = 0.0 site_10 1
# variable_mod_11 = 0.0 site_11 1
# variable_mod_12 = 0.0 site_12 1
# variable_mod_13 = 0.0 site_13 1
# variable_mod_14 = 0.0 site_14 1
# variable_mod_15 = 0.0 site_15 1
# variable_mod_16 = 0.0 site_16 1
allow_multiple_variable_mods_on_residue = 0
max_variable_mods_per_peptide = 3 # Maximum total number of variable modifications per peptide.
max_variable_mods_combinations = 5000 # Maximum number of modified forms allowed for each peptide (up to 65534).
output_format = pepXML_pin # File format of output files (tsv, pin, pepxml, tsv_pin, tsv_pepxml, pepxml_pin, or tsv_pepxml_pin).
output_report_topN = 1 # Reports top N PSMs per input spectrum.
output_max_expect = 50 # Suppresses reporting of PSM if top hit has expectation value greater than this threshold.
report_alternative_proteins = 1 # Report alternative proteins for peptides that are found in multiple proteins (0 for no, 1 for yes).
precursor_charge = 1 4 # Assumed range of potential precursor charge states. Only relevant when override_charge is set to 1.
override_charge = 0 # Ignores precursor charge and uses charge state specified in precursor_charge range (0 or 1).
digest_min_length = 7 # Minimum length of peptides to be generated during in-silico digestion.
digest_max_length = 50 # Maximum length of peptides to be generated during in-silico digestion.
digest_mass_range = 500.0 5000.0 # Mass range of peptides to be generated during in-silico digestion in Daltons.
max_fragment_charge = 2 # Maximum charge state for theoretical fragments to match (1-4).
track_zero_topN = 0 # Track top N unmodified peptide results separately from main results internally for boosting features.
zero_bin_accept_expect = 0 # Ranks a zero-bin hit above all non-zero-bin hit if it has expectation less than this value.
zero_bin_mult_expect = 1 # Multiplies expect value of PSMs in the zero-bin during results ordering (set to less than 1 for boosting).
check_spectral_files = 1 # Checking spectral files before searching.
minimum_peaks = 15 # Minimum number of peaks in experimental spectrum for matching.
use_topN_peaks = 150 # Pre-process experimental spectrum to only use top N peaks.
min_fragments_modelling = 2 # Minimum number of matched peaks in PSM for inclusion in statistical modeling.
min_matched_fragments = 4 # Minimum number of matched peaks for PSM to be reported.
min_sequence_matches = 2 # [nglycan/labile search_mode only] Minimum number of sequence-specific (not Y) ions to record a match.
minimum_ratio = 0.01 # Filters out all peaks in experimental spectrum less intense than this multiple of the base peak intensity.
clear_mz_range = 0.0 0.0 # Removes peaks in this m/z range prior to matching.
add_Cterm_peptide = 0.0
add_Nterm_peptide = 0.0
add_Cterm_protein = 0.0
add_Nterm_protein = 0.0
add_G_glycine = 0.0
add_A_alanine = 0.0
add_S_serine = 0.0
add_P_proline = 0.0
add_V_valine = 0.0
add_T_threonine = 0.0
add_C_cysteine = 57.02146
add_L_leucine = 0.0
add_I_isoleucine = 0.0
add_N_asparagine = 0.0
add_D_aspartic_acid = 0.0
add_Q_glutamine = 0.0
add_K_lysine = 0.0
add_E_glutamic_acid = 0.0
add_M_methionine = 0.0
add_H_histidine = 0.0
add_F_phenylalanine = 0.0
add_R_arginine = 0.0
add_Y_tyrosine = 0.0
add_W_tryptophan = 0.0
add_B_user_amino_acid = 0.0
add_J_user_amino_acid = 0.0
add_O_user_amino_acid = 0.0
add_U_user_amino_acid = 0.0
add_X_user_amino_acid = 0.0
add_Z_user_amino_acid = 0.0
Validate PSMs using PeptideProphet
Now that we have performed the database search, we can validate the identified PSMs using PeptideProphet and iProphet.
#!/bin/bash
mkdir -p library/tpp/
wdir=$(pwd)
# Run PeptideProphet on all pepXML files using all available CPU cores
ls -1 $wdir/library/msfragger/*_diatracer.pepXML | parallel -j+0 \
'input_file={}; \
base_name=$(basename "$input_file" .pepXML); \
log_file="library/tpp/peptideprophet_${base_name}.log"; \
echo "Processing $base_name - Output going to $log_file"; \
docker run --rm -v library/msfragger/:/msfragger/ -v library/tpp/:/tpp/ spctools/tpp:version7.20 \
# Note: We format and change the pepXML from MSFragger to be compatible with TPP using InteractParser, and use `pep.xml` as this is the expected file extension for downstream use in EasyPQP
bash -c "InteractParser /tpp/${base_name}.pep.xml /msfragger/${base_name}.pepXML && \
PeptideProphetParser /tpp/${base_name}.pep.xml ACCMASS NONPARAM DECOY=DECOY_ DECOYPROBS EXPECTSCORE PPM" > "$log_file" 2>&1'
# Run iProphet on all PeptideProphet pepXML files
docker run --rm -v library/tpp/:/tpp/ spctools/tpp:version7.2.0 \
bash -c 'InterProphetParser DECOY=DECOY_ /tpp/*.pep.xml /tpp/iprophet.pep.xml'
Generate Spectral Library using EasyPQP
We need to convert the pep.xml files to intermediate pickle files for EasyPQP. We can use the easypqp convert command to do this.
#!/bin/bash
mkdir -p library/easypqp/
wdir=$(pwd)
# Convert the psm tsv files to psm pickle and peak pickle files containing the spectral intensitiy information for EasyPQP
ls -1 $wdir/library/tpp/*_diatracer.pep.xml | parallel -j+0 \
'input_file={}; \
base_name=$(basename "$input_file" .pep.xml); \
log_file="easypqp_convert_${base_name}.log"; \
echo "Processing $base_name - Output going to $log_file"; \
easypqp convert --unimod unimod_phospho.xml --pepxml library/tpp/iprophet.pep.xml --psms library/easypqp/"$base_name".psmpkl --peaks library/easypqp/"$base_name".peakpkl --spectra library/msfragger/"$base_name"_calibrated.mzML > "$log_file" 2>&1'
Now we can generate the transition list using the easypqp library command.
#!/bin/bash
wdir=$(pwd)
cd library/easypqp/
# Run EasyPQP to generate the transition list from the psm and peak pickle files
easypqp library --out=easypqp_library.tsv --psm_fdr_threshold=0.01 --peptide_fdr_threshold=0.01 --protein_fdr_threshold=0.01 --rt_lowess_fraction=0.1 --pi0_lambda=0.05 0.5 0.05 --peptide_plot=easypqp_peptide_report.pdf --protein_plot=easypqp_protein_report.pdf *.psmpkl *.peakpkl > easypqp_library.log 2>&1
cd $wdir
We need to format the resulting transition list into the sqlite-lite based peptide query format (pqp) file used by OpenSwathWorkflow. We can use the OpenSwathAssayGenerator and OpenSwathDecoyGenerator commands to do this.
#!/bin/bash
docker run --rm -v $PWD/library/:/data/ ghcr.io/openms/openms-executables \
OpenSwathAssayGenerator -in /data/easypqp/easypqp_library.tsv -out /data/easypqp_library_targets.pqp -min_transitions 3 -max_transitions 6 -product_lower_mz_limit 300 -product_upper_mz_limit 1800 > /data/library/assay_generation.log 2>&1
docker run --rm -v $PWD:/data/ ghcr.io/openms/openms-executables \
OpenSwathDecoyGenerator -in /data/easypqp_library_targets.pqp -out /data/easypqp_library.pqp >> /data/library/assay_generation.log 2>&1
Generate Run-Specific iRT libraries
We can use the *_run_peaks.tsv files generated by easypqp to generate a run-specific iRTs for OpenSwathWorkflow. These will be used for retention time, mass-to-charge and ion mobility calibration during targeted data extraction. We will generate both nonlinear and linear iRT libraries.
#!/bin/bash
# Generate nonlinear iRTs
ls -1 library/easypqp/*_run_peaks.tsv | parallel -j+0 \
'input_file={}; \
base_name=$(basename "$input_file" _run_peaks.tsv); \
log_file="library/easypqp_irt_${base_name}.log"; \
echo "Processing $base_name - Output going to $log_file"; \
echo "Generating nonlinear iRTs for $base_name" >> "$log_file"; \
docker run --rm -v $PWD:/data/ ghcr.io/openms/openms-executables \
OpenSwathAssayGenerator -in /data/library/easypqp/"$base_name"_run_peaks.tsv -out /data/library/"$base_name"_irt_nonlinear.pqp >> "$log_file" 2>&1'
# Generate linear iRTs
ls -1 library/*_irt_nonlinear.pqp | parallel -j+0 \
'input_file={}; \
base_name=$(basename "$input_file" _irt_nonlinear.pqp); \
log_file="library/easypqp_irt_${base_name}.log"; \
echo "Processing $base_name - Output going to $log_file"; \
echo "Generating linear iRTs for $base_name" >> "$log_file" 2>&1; \
easypqp reduce --in library/"$base_name"_irt_nonlinear.pqp --out library/"$base_name"_irt_linear.pqp --bins 10 --peptides 20 >> "$log_file" 2>&1; \
echo "Done generating linear iRTs for $base_name" >> "$log_file" 2>&1;'
Targeted Data Extraction using OpenSwathWorkflow
Data Conversion
First, we need to convert the .d files to mzML format. We can use the diapysef tool converttdftomzml to do this.
Note
Ensure you have the Bruker SDK installed and properly configured on your system to enable diapysef to access and convert the .d files. You may need to run the diapysef converttdftomzml once so that the tool can try fetch the SDK, otherwise you can manually download it from here.
#!/bin/bash
# Run diapysef converttdftomzml on all .d directories using all available CPU cores
find . -maxdepth 1 -name "*.d" -type d | parallel -j+0 \
"echo 'Converting {} to {.}.mzML'; diapysef converttdftomzml --in={} --out={.}.mzML"
Converting ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740.d to ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740.mzML
Found Bruker sdk. Access to the raw data is possible.
You would expect output similar to the following:
[2025-09-29 22:10:41] INFO: Converting ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740.d...
Analysis has 3136 frames.
[2025-09-29 22:16:22] INFO: Finished converting TDF data to mzML!
100%|██████████| 3136/3136 [05:40<00:00, 9.22it/s]
Running OpenSwathWorkflow
Now we have all the required input components to run the OpenSwathWorkflow command to perform targeted data extraction.
#!/bin/bash
mkdir -p tmp/
mkdir -p openswath/
mkdir -p openswath/calibration/
# Using GNU Parallel for better job control
find . -maxdepth 1 -name "*.mzML" -type f | parallel -j1 \
'input_file={}; \
base_name=$(basename "$input_file" .mzML); \
log_file="openswath/openswath_${base_name}.log"; \
echo "Processing $base_name - Output going to $log_file"; \
# Create a temporary directory for intermediate files for each run
mkdir -p tmp/$base_name; \
# Run OpenSwathWorkflow
docker run --rm -v $PWD:/data/ ghcr.io/openms/openms-executables \
OpenSwathWorkflow \
-in /data/"$base_name".mzML \
-tr /data/library/easypqp_library.pqp \
-tr_irt /data/library/"$base_name"_diatracer_irt_linear.pqp \
-tr_irt_nonlinear /data/library/"$base_name"_diatracer_irt_nonlinear.pqp \
-out_features /data/openswath/"$base_name".osw \
-out_chrom /data/openswath/"$base_name".sqMass \
-readOptions cacheWorkingInMemory \
-tempDirectory /data/tmp/$base_name/ \
-batchSize 1000 \
-pasef \
-rt_extraction_window 250 \
-extra_rt_extraction_window 150 \
-mz_extraction_window 25 \
-mz_extraction_window_unit ppm \
-mz_extraction_window_ms1 25 \
-mz_extraction_window_ms1_unit ppm \
-ion_mobility_window 0.06 \
-im_extraction_window_ms1 0.06 \
-irt_mz_extraction_window 40 \
-irt_mz_extraction_window_unit ppm \
-irt_im_extraction_window 99 \
-min_coverage 0.6 \
-min_rsq 0.95 \
-min_upper_edge_dist 1 \
-ms1_isotopes 3 \
-mz_correction_function quadratic_regression_delta_ppm \
-Debugging:irt_trafo /data/openswath/calibration/"$base_name"_debug_calibration_irt.trafoXML \
-Debugging:irt_mzml /data/openswath/calibration/"$base_name"_debug_calibration_irt_chrom.mzML \
-Calibration:debug_mz_file /data/openswath/calibration/"$base_name"_debug_calibration_mz.txt \
-Calibration:debug_im_file /data/openswath/calibration/"$base_name"_debug_calibration_im.txt \
-RTNormalization:estimateBestPeptides \
-RTNormalization:alignmentMethod lowess \
-RTNormalization:lowess:span 0.01 \
-Scoring:Scores:use_ion_mobility_scores \
-threads 8 -outer_loop_threads 15 -force > "$log_file" 2>&1; \
# Clean up temporary directory
rm -rf tmp/$base_name'
Since we added addition outputs for debugging the calibration, we can inspect the calibration results using the pyprophet export calibration-report command.
#!/bin/bash
wd=$PWD
cd openswath/calibration/
# Generate calibration report
pyprophet export calibration-report
cd $wd
We would expect output similar to the following:
PyProphet v3.0.0
Execution time: 2025-09-30 12:56:41
System: OS: Linux 6.9.3-76060903-generic | Python: 3.10.14 | CPU: 20 cores | RAM: 62.4 GB
Command: /home/singjc/anaconda3/envs/py310/bin/python /home/singjc/anaconda3/envs/py310/bin/pyprophet export calibration-report
INFO: Found 3 unique runs to generate reports for
---------------------------------------------------------------------
INFO: Processing run - 20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A1_1_2737
INFO: IM calibration file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A1_1_2737_debug_calibration_im.txt
INFO: MZ calibration file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A1_1_2737_debug_calibration_mz.txt
INFO: iRT transformation file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A1_1_2737_debug_calibration_irt.trafoXML
INFO: iRT XIC mzML - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A1_1_2737_debug_calibration_irt_chrom.mzML
INFO: Zooming into XICs if possible...
---------------------------------------------------------------------
INFO: Processing run - 20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A2_1_2738
INFO: IM calibration file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A2_1_2738_debug_calibration_im.txt
INFO: MZ calibration file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A2_1_2738_debug_calibration_mz.txt
INFO: iRT transformation file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A2_1_2738_debug_calibration_irt.trafoXML
INFO: iRT XIC mzML - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A2_1_2738_debug_calibration_irt_chrom.mzML
INFO: Zooming into XICs if possible...
---------------------------------------------------------------------
INFO: Processing run - 20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740
INFO: IM calibration file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740_debug_calibration_im.txt
INFO: MZ calibration file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740_debug_calibration_mz.txt
INFO: iRT transformation file - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740_debug_calibration_irt.trafoXML
INFO: iRT XIC mzML - ./20200505_Evosep_200SPD_SG06-16_MLHeLa_200ng_py8_S3-A4_1_2740_debug_calibration_irt_chrom.mzML
INFO: Zooming into XICs if possible...
INFO: Done. Saved report to 'calibration_report.pdf'. Elapsed: 0 days 00:00:11.934842
[ 2025-09-30 at 12:56:53 | INFO | util::543 ] pyprophet export_calibration_report took 11.96 seconds; Peak Memory Usage: 433.62 MB.
The resulting calibration_report.pdf file contains scatter plots showing the correlation between the observed and expected values for retention time (iRT), mass-to-charge ratio (m/z), and ion mobility (IM). Additionally, it includes a few sample extracted ion chromatograms (XICs) for the iRT peptides used in the calibration process.
Statistical Validation
Now that we have identified peak-group features in the data, we can perform semi-supoervised learning to score the features and compute false discovery rate (FDR) estimates using PyProphet.
# Score the peak-group features
pyprophet score --in=pyprophet/merged.osw --level=ms1ms2 --classifier=XGBoost --ss_num_iter=3 --xeval_num_iter=3 --threads=3 --ss_scale_features
# Compute peptide and protein level FDR estimates
pyprophet infer peptide --in=pyprophet/merged.osw --context=global
pyprophet infer peptide --in=pyprophet/merged.osw --context=experiment-wide
pyprophet infer protein --in=pyprophet/merged.osw --context=global
pyprophet infer protein --in=pyprophet/merged.osw --context=experiment-wide
Exporting results
Finally, we can export the results to a tsv file for downstream analysis.
# Export the results to a tsv file, where each row is a precursor peak-group feature in a run
pyprophet export tsv --in=pyprophet/merged.osw --out=pyprophet/merged.tsv
# You can optionally export quantification matrices as well
pyprophet export matrix --in=pyprophet/merged.osw --out=pyprophet/peptide_matrix.tsv --level peptide
pyprophet export matrix --in=pyprophet/merged.osw --out=pyprophet/protein_matrix.tsv --level protein
You can export a pdf that compiles key scoring diagnostics and across-run QC: target/decoy d-score histograms and densities, q-/s-value curves, p-value/π₀ fit, IDs per run, run-intersection curves, Jaccard similarity, intensity correlations, CV and violin plots, plus summary tables at 1/5/10% FDR across precursor/peptide/protein levels (including IPF when available). Use it for a quick, high-level check of identification quality, FDR behavior, and quant reproducibility across runs.
pyprophet export score-report --in merged.osw