Basic usage

Always make sure that you have run your database search with 100% protein-level FDR. Instructions on how to do this are given in the respective search engine sections.

Users unfamiliar with the command line and Python are advised to use the Graphical user interface (GUI). For more flexibility, use the Command line or call functions directly in Python.

Graphical user interface (GUI)

Starting the GUI

On Windows (no installation required):

Download and unzip the PickedGroupFDR_GUI_windows.zip from the latest release.
Double click on PickedGroupFDR.exe to start the GUI. It may take a few seconds for the GUI to open.

On all platforms (Windows, Linux, OSX, etc.):

Clone and install this repository

git clone https://github.com/kusterlab/picked_group_fdr.git
cd picked_group_fdr
pip install .

Install PyQt5 with pip:
pip install PyQt5
Start the GUI from the repo directory with:
python gui.py

Using the GUI

The example below shows how to analyze the data downloadable from https://zenodo.org/records/10056025.

Select the evidence files.
If necessary, update the digestion parameters in the table.
Select the fasta file(s).
If using rescoring results from Oktoberfest
1. check the Use rescoring results checkbox.
2. Add the rescored PSM files generated by Oktoberfest.
Keep the Do quantification checkbox checked
Depending on your preference, set LFQ min peptides to 1 or 2. This is the minimum number of peptides for a protein to be quantified.
Keep the Protein group-level FDR cutoff at 0.01, which corresponds to a 1% protein group-level FDR for the final proteinGroups.txt.
Specify the output directory.
Click on Run.
The filtered protein group results can be found in the output directory specified in Step 8 in a file called proteinGroups.fdr1.txt.

Command line

MaxQuant results

Install the Picked Group FDR Python package.
Make sure that you have run the MaxQuant search with 100% protein-level FDR.

The posterior error probabilities (PEP) of MaxQuant are not well-calibrated. Therefore, we first recalculate these with Mokapot (=Percolator for Python):

python3 -u -m picked_group_fdr.pipeline.andromeda2pin </path/to/mq_evidence_txt> \
   --outputTab andromeda.tab \
   --databases </path/to/fasta_file>
python3 -u -m picked_group_fdr.pipeline.run_mokapot 0.01 0.01 percolator <num_threads>
python3 -u -m picked_group_fdr.pipeline.update_evidence_from_pout \
   --mq_evidence </path/to/mq_evidence_txt> \
   --perc_results percolator/andromeda.mokapot.psms.txt percolator/andromeda.mokapot.decoy.psms.txt \
   --mq_evidence_out percolator/evidence.txt

Alternatively, you can use Prosit’s Percolator results files directly:

python3 -u -m picked_group_fdr.pipeline.update_evidence_from_pout \
   --mq_evidence </path/to/mq_evidence_txt> \
   --perc_results prosit_target.psms prosit_decoy.psms \
   --mq_evidence_out percolator/evidence.txt \
   --pout_input_type prosit

To obtain protein group level FDRs, run:

python -m picked_group_fdr \
   --mq_evidence percolator/evidence.txt \
   --fasta </path/to/fasta_file> \
   --method picked_protein_group_mq_input \
   --protein_groups_out percolator/proteinGroups.txt

Python

Protein inference

To only run the protein inference functionality on a list of identified peptides with associated proteins, call the picked_group_fdr.get_protein_group_results function on a dictionary of peptide_sequence => (peptide_posterior_error_prob, [protein1, ...]):

from picked_group_fdr import picked_group_fdr

results = picked_group_fdr.get_protein_group_results(
   {"PEPA": (1e-5, ["PROTA", "PROTB"])}
)

print(results.protein_group_results)
# prints [ProteinGroupResult(proteinIds='PROTB;PROTA', majorityProteinIds='PROTB;PROTA',
#         peptideCountsUnique='1;1', bestPeptide='PEPA', numberOfProteins=2, qValue=0.5,
#         score=5.0, reverse='', potentialContaminant='', precursorQuants=[], extraColumns=[])]

This functionality is also demonstrated in a more complete manner in a Jupyter notebook available at https://github.com/kusterlab/picked_group_fdr/tree/main/data/protein_inference_example/protein_inference_and_fdr.ipynb

Python pipeline / jupyter notebook

The PickedGroupFDR Python module exposes a number of convenient methods for calling the different tools inside a Python script. Here, this functionality is demonstrated using a Jupyter notebook available at https://github.com/kusterlab/picked_group_fdr/tree/main/data/book_chapter/coon_analysis.ipynb:

The example below shows how to analyze the data downloadable from https://zenodo.org/records/10056025.

Import the pipeline module of the PickedGroupFDR package, as well as the DigestionParams class. The pipeline module contains several methods for calling the different tools in PickedGroupFDR. The DigestionParams class provides a wrapper for all digestion parameters.
If using rescoring results from Oktoberfest, update the MaxQuant evidence files with the rescored PSMs from Oktoberfest using the run_update_evidence method.
Process the (updated) evidence files with PickedGroupFDR using the run_picked_group_fdr method.
Filter the results at 1% FDR using the run_filter_fdr_maxquant method.
Open the filtered results with the Python pandas package.