STAN Community Benchmark

Community Reference Ranges

Expected performance ranges established by the community. Use the filters below to find your instrument and method. Ranges update automatically as more labs contribute.

Loading community data...

Best Configurations

Top instrument × throughput × amount-loaded combinations from community data. Each row is one cohort (≥3 submissions). Click any column header to re-rank — pick your priority (depth or accuracy) and read the row.

Instrument Health Explorer

Each point represents an anonymous submission. Distributions show the community range for each instrument family. Hover for details.

Depth by Amount Loaded

How does sample load affect identification depth? Each violin is one amount bucket, colored by instrument model. Saturation typically appears between 100–250 ng on Orbitrap; timsTOF often climbs further. Sample type from the page filter above; DIA/DDA from the tab below.

Identification Depth by Platform

How many precursors (DIA) or PSMs (DDA) does each platform typically identify? Apples-to-apples comparison — pick a load amount so 5 ng K562 isn't pooled with 200 ng HeLa. Each violin is one instrument model (Pro vs Pro 2 vs HT are separate). Sample type comes from the page-level filter above.

Amount loaded: Shape: ○ 50 ng · ◇ <20 ng · □ >100 ng

Depth by Throughput (which SPD gives me the best data?)

Precursors/PSMs vs. samples-per-day, faceted by instrument family. Each box is one SPD bucket. Filter by amount loaded so you compare apples to apples — the optimal SPD may differ by loading.

Amount loaded: Shape: ○ 50 ng · ◇ <20 ng · □ >100 ng

Identification Depth vs. IPS

Cohort-normalized depth score (IPS) vs. precursor count. IPS is a depth score — not an LC-health metric. Shape: ● timsTOF HT · ◆ Astral · ■ Exploris 480 · ▲ Lumos. Color = SPD. Size ∝ peptide count.

Column Comparison (same instrument, SPD, and amount)

How does your LC column compare to others under identical conditions? Each group shows submissions matched by instrument family, SPD, and injection amount — only the column differs.

Throughput vs. Quantitation Quality (Matthews & Hayes 1976)

SPD vs. data points across peak. Below 6 points, quantitation error exceeds 1%. Shape = LC column. Color = instrument family.

LC / Instrument Health (ID-free metrics)

These are the metrics the 2024 proteomics QC literature (NIST MSQC, QCloud2, CPTAC, PTXQC) considers the real LC-health signals — they catch failures BEFORE identifications collapse. Unlike IPS (which is a depth rank), these don't depend on how many peptides you identified.

Mass Accuracy Drift (MS1)

Median corrected MS1 mass error per run. Tracks Orbitrap / qTOF calibration stability. A rising trend signals lock-mass failure or thermal drift before IDs drop.

MS1 Signal (TIC proxy)

Total MS1 ion current per run — a proxy for ion source health. Drops of >2× signal dirty emitter, low flow, or sample prep issue before IDs fall.

Dynamic Range

log₁₀(p99 / p01) of precursor intensity. Compresses when the ion source is dirty or the LC is losing pressure. Populated going forward by the STAN watcher.

Points Across Peak

Datapoints per chromatographic peak (per Matthews & Hayes 1976). A rising trend at constant SPD signals column degradation (peaks broadening). Validated against Spectronaut (median 9 on timsTOF 100 SPD).

Instrument Health Fingerprint

Each polygon is one (instrument model × SPD tier) cohort with ≥3 DIA submissions, plotting the median across that cohort. Axes: 3 depth metrics + MS1 mass accuracy (lower ppm → better → outer ring). Normalization is relative to the cohort-median range, so the smallest cohort pins to the center and the largest to the edge.

Community TIC Overlay by SPD

Identified (DIA) or raw (DDA) TIC chromatograms grouped by throughput + LC system + acquisition mode. For Evosep users the gradient is standardized — shape differences reveal instrument-specific issues. Thick dashed line = community median. Always pick one acquisition mode — DIA and DDA have different cycle times and their shapes should not be averaged together. First-of-its-kind cross-lab TIC comparison.

Your Lab vs. Community

Your instrument's QC trend over time, with community reference bands (mean ± 1/2/3σ from all submissions of the same instrument family). Use this to answer: "Is my instrument drifting, and am I above or below the community baseline?"

Lab: Instrument: Metric: Amount:

Your Trend vs. Community Reference

Blue dots = your lab's runs over time. Shaded bands = community mean ± 1σ (green), ± 2σ (amber), ± 3σ (red). Runs outside ±2σ deserve a look; outside ±3σ = something is wrong or exceptional.

Understanding the Metrics

The HeLa Standard

All STAN benchmarking uses the Pierce HeLa Protein Digest Standard (Thermo Scientific, cat# 88328 / 88329). This is a tryptic digest of the HeLa S3 cell line containing >15,000 proteins with <10% missed cleavages.

Using a single, commercially available standard ensures every lab starts from the same sample. Differences in metrics reflect instrument and LC performance, not sample preparation variability.

Buy 20 μg (88328) · Buy 5 x 20 μg (88329)

IPS: Instrument Performance Score (0-100)

A composite score computed entirely from search output — no reference run, no blanks, works from the very first QC injection.

DIA: 30% precursor depth · 25% spectral quality (frags/precursor) · 20% sampling (pts/peak) · 15% quant coverage · 10% digestion
DDA: 30% PSM depth · 25% mass accuracy · 20% sampling (pts/peak) · 15% hyperscore · 10% digestion

90-100 Excellent 80-89 Good 60-79 Marginal <60 Investigate

Why This Benchmark Works

Every community submission searches the exact same frozen, hash-verified human UniProt FASTA and predicted spectral library. Identical upstream parameters mean differences in output reflect instrument performance, not search configuration.

Primary: Precursors (DIA) / PSMs (DDA) — purest instrument signal
Secondary: Peptides — slight sensitivity to search settings
Context: Proteins — shown for reference, not used for ranking (still affected by inference algorithm)
Health: IPS, missed cleavages, charge distribution

Points Across Peak

The number of MS2 scans sampling each precursor's elution profile directly determines quantitation accuracy (Matthews & Hayes, 1976).

12+ points: reliable quantitation
6-12 points: minimum for acceptable accuracy (depends on peak shape)
<6 points: systematic quantitation error increases rapidly

At high SPD with short columns, cycle time can exceed peak width. These are guidelines — actual error depends on peak symmetry and integration method. Track this metric to find the throughput limit for your setup.

Community Submissions

All submissions are anonymous by default. Click column headers to sort. Use the export button to download as CSV.

Loading submissions...