LFQ, precursor ion (Astral, low input)#

alpha Quantification — DIA, Orbitrap Astral, low input (200 pg)

This module is in alpha stage

This module is still in active development. Its scope, metrics, or dataset may change as we refine it based on community feedback. Submissions are welcome, but please be aware that comparisons may shift in future versions.

Compares the sensitivity and quantification accuracy of workflows for data-independent acquisition (DIA) data, namely narrow-window 2 Th, on an Orbitrap Astral (Thermo Fisher Scientific), on low-input samples of approximately 200 pg protein amount.

For bulk sample analyses on the same instrument type, see LFQ, precursor ion (Astral); for the equivalent low-input dataset on a timsTOF, see LFQ, precursor ion (timsTOF, low input).

Open in web app

At a glance#

Acquisition

DIA, 2 Th windows

Instrument

Orbitrap Astral, low input (~200 pg)

Level

Precursor ion (modified sequence + charge)

Metric

Epsilon (quantification accuracy)

What this module tests#

This module is best suited to evaluate the impact of, among others:

  • search engine identification

  • peak picking

  • low-level ion signal normalization

It is not designed to evaluate later-stage post-processing of quantitative data (e.g. missing value replacement or manual filtering); please upload data without that kind of post-processing. Other modules focus specifically on those steps.

Dataset#

An Astral dataset with low-input three-proteome mixture samples, with two distinct abundance ratios of tryptic digest, each injected at a total of 200 pg protein amount. This corresponds to logarithmic fold changes (log2FC) of 0, −2, and 1 for human, E. coli, and yeast respectively. The human proportion was sourced from Pierce™ HeLa Protein Digest Standard (Thermo Fisher Scientific, #88328).

Chromatographic separation used a Vanquish Neo UHPLC system (Thermo Fisher Scientific) with a two-column setup: a PepMap C18 trapping column (5 mm × 300 μm) for online desalting, coupled to an IonOpticks Aurora Series C18 analytical column (8 cm × 75 μm). The chromatographic gradient was initiated at 4% B (20% H2O, 79.9% acetonitrile, 0.1% FA) and a flow rate of 800 nL, ramped to 12% B within 0.075 min, followed by 18% within 0.188 min, 35% in 0.8 min and 99% within 0.05 min. Held at 99% for 0.338 min, flow rate was decreased to 100 nL and held for 6.249 min afterwards increased to 800 nL and 4% B in 0.05 min.

Acquisition on the Orbitrap Astral ran for 7.8 min in Peptide mode: spray voltage 1.9 kV, 240 °C ion transfer tube temperature, FAIMS standard resolution (3.6 L/min carrier gas, CV −48 V). Full scans covered 400–800 m/z at 240,000 resolution (45% RF lens, 500% AGC target, 100 ms max injection time). DIA windows were dynamically placed starting from 400-800 m/z (400-437, 437-454, 454- 472, 472-485, 485-498, 498-511, 511-523, 523-536, 536-547, 547-599, 599-572, 572- 584, 584-596, 596-611, 611-626, 626-645, 645-664, 664-688, 688-722, 722-800 m/z). HCD collision energy was set to 25%, ions were collected until an AGC target of 500% (absolute value of 5x104 ) or maximum injection time of 40 ms was reached.

Note

Raw files are not yet on ProteomeXchange; the reserved identifier is PXD080610.

Download:

Important

Do not rename the downloaded raw files.

How the metric is calculated#

For each precursor ion, ProteoBench sums the signal per raw file, removes contaminants (flagged Cont_ in the FASTA) and precursors matching multiple species, log2-transforms the remaining values, then computes the mean and coefficient of variation (CV) per condition. The difference between the mean log2 intensity in A and B is compared against the expected log2 fold change for that precursor’s species — that difference is epsilon.

The main plot shows the number of unique precursor ions quantified (vertical axis) against the mean or median absolute epsilon (horizontal axis).

Run your workflow#

Run your analysis on the downloaded files, then follow Your First Submission to upload, inspect, and (if you’d like) submit your results publicly.

Tool-specific setup#

Table: input files required for metric calculation and public submission

Tool

Input file

Parameter file

AlphaDIA

precursors.parquet/.tsv (v2+) or precursor.matrix.tsv + precursors.tsv (v1)

log_alphadia.txt

Custom

custom_input.tsv

—

DIA-NN

report.tsv or report.parquet

report.log.txt

FragPipe

combined_ion.tsv

fragpipe.workflow

FragPipe (DIA-NN quant)

report.tsv or report.parquet

fragpipe.workflow

MSAID

MSAID_output.tsv

MSAID_params.csv

MaxQuant

evidence.txt

mqpar.xml

PEAKS

lfq-features.csv

*.txt

Spectronaut

*.tsv

ExperimentSetupOverview.txt

Expand a tool below for setup details.

DIA-NN
  1. Import raw files.

  2. Add the FASTA, but do not select “Contaminants” (already included).

  3. Turn on FASTA digest for library-free search / library generation.

  4. Do not set verbosity/log level above 1, or parameter parsing will fail.

  5. Upload *_report.tsv or *_report.parquet for scoring and report.log.txt for public submission.

AlphaDIA
  1. Select the FASTA and import .raw files in “Input files”.

  2. Define search parameters in “Method settings”.

  3. Turn on “Predict Library”.

  4. Turn on “Precursor Level LFQ”.

  5. Required output files depend on version: AlphaDIA v1.x needs both precursors.tsv and precursor.matrix.tsv (or use the conversion notebook); later versions need only precursors.parquet/.tsv.

FragPipe (DIA-NN quant)
  1. Load the DIA_SpecLib_Quant workflow.

  2. After importing raw files, assign experiments “by File Name”.

  3. Do not add contaminants when adding decoys to the database.

  4. Upload *report.tsv for scoring and fragpipe.workflow for public submission.

FragPipe reports protein identifiers across two columns (“Proteins” and “Mapped Proteins”); these are concatenated to form the protein groups.

Spectronaut

Accepted format: BGS Factory Report — ..._Report.tsv for scoring, ..._Report.setup.txt for parameter parsing.

  1. Import the module FASTA in “Databases” (UniProt parsing rule).

  2. In “Analysis”, select “Set up a DirectDIA Analysis from folder” and load the raw-file folder.

  3. Select the FASTA as the database.

  4. Choose search settings.

  5. Assign conditions: ..._Condition_A_REP1–REP3 to “A”, ..._Condition_B_REP1–REP3 to “B”.

  6. Skip GO terms/library extensions.

  7. Run the search, then export a BGS Factory Report as ..._Report.tsv.

  8. Upload ..._Report.tsv privately and ..._Report.setup.txt for public submission.

MaxDIA (work in progress)

By default MaxDIA uses its own contaminants-only FASTA. This module’s FASTA already includes a curated contaminant set, so untick “Include contaminants” (Global parameters → Sequences), and set FASTA parsing to Identifier rule = >([^\t]*), Description rule = >(.*). Use “No Fractions” and name experiments A_REP1–A_REP3, B_REP1–B_REP3.

Upload evidence.txt for scoring and mqpar.xml for public submission.

PEAKS

Rename samples (Sample 1→6) to match the raw file names exactly: LFQ_Astral_DIA_15min_50ng_Condition_A_REP1–REP3, ..._Condition_B_REP1–REP3.

ProteoBench automatically corrects small naming differences (case, spacing, minor typos) in PEAKS sample names, but anything it can’t confidently resolve will still fail with a message telling you which name(s) didn’t match. Exact names are still the safest bet.

Set Enzyme = trypsin, Instrument = Orbitrap (Astral), Fragment = HCD, Acquisition = DIA. In the workflow, use the Quantification option; define search parameters in “DB search”. In “Quantification” use “Label Free” (individually or grouped by condition); in “Report” set both Precursor/Peptide FDR and Protein Group FDR to 1%. Check “All Search Parameters” and the “Feature Vector CSV” (Export tab) once finished.

Custom format

If your tool isn’t listed above, upload a tab-delimited table with:

  • Sequence — unmodified peptide sequence

  • Proteins — ;-separated identifiers, including the species flag (e.g. _YEAST)

  • Charge — precursor charge

  • Modified sequence — sequence with localized modifications, in ProForma

  • one quantitative column per sample: LFQ_Astral_DIA_15min_50ng_Condition_A_REP1 … LFQ_Astral_DIA_15min_50ng_Condition_B_REP3

The table must not contain non-validated ions.

How ProteoBench maps each tool’s columns#

Each tool’s output format is described in a .toml file under proteobench/io/parsing/io_parse_settings/, defining [mapper] (column-name mapping), [condition_mapper]/[run_mapper] (raw file to condition/sample), [species_mapper] (protein-ID suffix to species), [general] (contaminant flag Cont_ and decoy flag), and [modifications_parser] where needed.

Result columns#

After upload, the results table includes: the precursor ion (modified sequence + charge); mean and standard deviation of log2-transformed and of raw intensity per condition; CV per condition; the difference of mean log2 values between conditions; per-raw-file intensity; the number of raw files with a non-missing value; species and whether the sequence is species-specific; the expected ratio for that species; and epsilon.

Use the slider to set the minimum number of raw files a precursor must be quantified in to be included in the plot.

Parameters tracked for public submission#

Upload your parameter file under “Download calculated ratios”; see Tool-specific setup above for which file that is per tool. ProteoBench tracks:

  • software tool name and version; search engine name and version, if different

  • FDR threshold (PSM, precursor, peptide, protein level)

  • match-between-runs (on/off)

  • precursor and fragment m/z range and mass tolerance

  • enzyme (Trypsin, for this dataset) and maximum missed cleavages

  • minimum/maximum peptide length

  • fixed and variable modifications, and the maximum number of modifications

  • minimum/maximum precursor charge

If any of these are missing from your parameter file, add them in the Comments for submission field.

Check your parameter file for personal information

Parameter files can embed local file paths (FASTA location, raw data location, tool installation paths) that may reveal personal or institutional information. Review and sanitize before submission.

Once submitted you’ll get a pull-request link; save it to track your submission (see what happens next). Contact us or open an issue with any problems.