TEAPOT

Targeted Evaluation & Analysis Pipeline On Tandem mass spectrometry

TEAPOT is an analysis pipeline for large-scale MS-based targeted proteomics, on both PRM and DIA data. It scales to cohorts of thousands of runs, processed in a single automated pass, reproducibly, on a local machine or an HPC cluster.

Every peptide is assigned a $q$ value and a posterior error probability (PEP), and abundances are corrected for interference. Label-free runs give relative abundances; SIL internal standards of known amount give absolute abundances. Quantification accuracy improves as more runs are added to a cohort.

Highlights

• No manual work. Targeted PRM is typically analyzed manually in Skyline. TEAPOT runs PRM and DIA analyses end to end, in parallel, at cohort scale.
• From instrument files to results. Which peptides were detected, at what confidence, and how much of each. Nothing for you to prepare in advance or hand in between.
• Bring whatever you have. An existing library (.elib, .blib or .dlib), Gas-Phase Fractionation (GPF) runs, or just a FASTA. TEAPOT puts it into the form the search needs.
• Confidence on every peptide. A $q$ value and a PEP for every identification.
• Abundances corrected for interferences. Peptides are quantified using evidence from all runs in the cohort, with interferences separated from their real signal.

---

Does TEAPOT fit your data?

you have	you get
PRM runs, and a reference list of the peptides you care about	which of them were detected at a 1 % $q$ value, and how much of each
PRM runs, plus SIL internal standards of known amount	the same, and an absolute amount in pmol
DIA runs	the same, across the whole assay library instead of a reference list

How TEAPOT works

Reference and background peptides

Two terms run through the input templates and the rest of this page.

Reference peptides are the ones you want quantities for: in a label-free run the peptides the mass spectrometer was scheduled to acquire, in a SIL run the heavy internal standards and their endogenous light counterparts. Background peptides are the other peptides present in the sample that you do not need quantities for.

Context and Context-MS need both to estimate confidence, so your library has to cover both. Most libraries do, but a vendor .blib of SIL internal standards is reference-only. With such a library you need to supply a background if you search it with EncyclopeDIA on PRM data, or with OpenSWATH on PRM data using Context-MS.

Quick start

Requirements

• Nextflow 23.04 or newer, and Java 17 or newer
• Apptainer

Which workflow to use

your data	recommended workflow	alternative
Label-free targeted PRM	encyclopedia_prm_label_free	openswath_prm_label_free
Targeted PRM with SIL internal standards, absolute quant	encyclopedia_dia	encyclopedia_prm_sil_standards, openswath_prm_sil_standards
Wide-window DIA	encyclopedia_dia	openswath_dia

Then set up a run from it: copy the workflow's templates into your own working directory and fill in every placeholder:

• its .csv samplesheet and .yml of settings, from templates/
• a launcher, run_local.sh for one local machine or run_slurm.sh for an HPC cluster

On an HPC cluster, two more edits: your SLURM account in conf/slurm.config, and your cluster's Java and Nextflow module names in run_slurm.sh.

What you need to supply

• A samplesheet CSV. One row per run, with the header sample_id,file,mode,reference_list,engine:

  • mode is PRM or DIA
  • engine is encyclopedia or openswath
  • reference_list is your list of reference peptides, and can be left empty when the pipeline can derive it (see below)

• Run files, named in the samplesheet's file column. Either an open format (.mzML, .mzXML) or a vendor format the pipeline converts for you: .raw (Thermo), .wiff (SCIEX), .d (Bruker/Agilent), .lcd (Shimadzu). A .wiff is not a single file. Name only the .wiff and keep its paired .wiff.scan in the same directory, and the pipeline picks it up automatically.

• An assay library, as any one of:

  • a library you already have: .elib, .blib or .dlib, or with OpenSWATH a .pqp, transition .tsv or TraML
  • Gas-Phase Fractionation (GPF) runs, listed in a CSV as in templates/library_sheet.csv, and the pipeline builds a chromatogram library (.elib) from them
  • nothing, and the pipeline predicts one from your FASTA

• Sequences in FASTA. For label-free PRM and DIA runs, the proteins potentially present in your samples. For SIL runs, the standard construct sequences supplied with the vendor kit.

• A background, if you need one (see Reference and background peptides). Either a library (.elib, .blib or .dlib) or a FASTA of background proteins.

• A reference list, for PRM runs. Either name it in the samplesheet's reference_list column, or supply a vendor .blib and the pipeline derives one from it. This is typically the list you gave the mass spectrometer when setting up the acquisition. DIA runs do not need one.

• Known amounts, only for absolute quantification. The pmol of each SIL internal standard, usually supplied with the vendor kit and read as-is by the pipeline.

INPUTS.md has the layout and requirements of every input.

Commands

# 1. get the pipeline
git clone https://github.com/zyqqqq0122/teapot.git
TEAPOT_DIR=$PWD/teapot

# 2. work from a directory of your own, not the pipeline checkout
mkdir -p ~/my_analysis && cd ~/my_analysis

# 3. copy the templates that matches your data, from the table above
cp $TEAPOT_DIR/templates/encyclopedia_prm_label_free.yml my_run.yml
cp $TEAPOT_DIR/templates/encyclopedia_prm_label_free.csv my_samplesheet.csv
cp $TEAPOT_DIR/templates/run_slurm.sh .        # or cp $TEAPOT_DIR/templates/run_local.sh . on a local machine

# 4. edit the three files as described above

# 5. run it
sbatch run_slurm.sh                            # or ./run_local.sh on a local machine

How to read the results

results/
  quant/<route>/
    quant.reference.tsv          one row per reference peptide per run: found or not, and how much
    peptide.pairs.confident.tsv  light/heavy pairs and absolute pmol, SIL runs only
    ...                          the full table, the confident subset, and a per-run summary
  pipeline_info/                 logs, timings, software versions, the Nextflow report
  ...                            intermediates from every step

<route> names the engine and acquisition mode. For example, enc_prm is EncyclopeDIA on PRM runs.

OUTPUTS.md describes every file and column.

Getting help

• INPUTS.md: every input file, its layout and accepted columns
• OUTPUTS.md: every file in results/, and every column in it
• PARAMETERS.md: every parameter, its default and when to change it
• ARCHITECTURE.md: how the pipeline is built, for developers

License

MIT

References

Software

• Context
• Context-MS
• diathem
• OpenMS
• PyProphet
• msproteomicstools
• Koina
• Nextflow
• ProteoWizard

Publications

• EncyclopeDIA. Searle BC, et al. Chromatogram libraries improve peptide detection and quantification by data independent acquisition mass spectrometry. Nat Commun 9, 5128 (2018). doi:10.1038/s41467-018-07454-w
• OpenSWATH. Röst HL, et al. OpenSWATH enables automated, targeted analysis of data-independent acquisition MS data. Nat Biotechnol 32, 219–223 (2014). doi:10.1038/nbt.2841
• PyProphet. Rosenberger G, et al. Statistical control of peptide and protein error rates in large-scale targeted data-independent acquisition analyses. Nat Methods 14, 921–927 (2017). doi:10.1038/nmeth.4398
• mProphet. Reiter L, et al. mProphet: automated data processing and statistical validation for large-scale SRM experiments. Nat Methods 8, 430–435 (2011). doi:10.1038/nmeth.1584
• Percolator. Käll L, et al. Semi-supervised learning for peptide identification from shotgun proteomics datasets. Nat Methods 4, 923–925 (2007). doi:10.1038/nmeth1113
• TRIC. Röst HL, et al. TRIC: an automated alignment strategy for reproducible protein quantification in targeted proteomics. Nat Methods 13, 777–783 (2016). doi:10.1038/nmeth.3954
• Prosit. Gessulat S, et al. Prosit: proteome-wide prediction of peptide tandem mass spectra by deep learning. Nat Methods 16, 509–518 (2019). doi:10.1038/s41592-019-0426-7
• Koina. Lautenbacher L, et al. Koina: Democratizing machine learning for proteomics research. Nat Commun 16, 9933 (2025). doi:10.1038/s41467-025-64870-5
• Nextflow. Di Tommaso P, et al. Nextflow enables reproducible computational workflows. Nat Biotechnol 35, 316–319 (2017). doi:10.1038/nbt.3820
• ProteoWizard. Chambers MC, et al. A cross-platform toolkit for mass spectrometry and proteomics. Nat Biotechnol 30, 918–920 (2012). doi:10.1038/nbt.2377