Home

The data-driven peptide search engine and quantification engine of the quantms ecosystem. Spectra and a FASTA go in; Percolator-ready PSMs and, on request, protein groups and label-free or TMT quantities come out. The scoring model is chosen per file from its metadata.

Get started Benchmarks Parameters

What it does

• Reads your files as they are

  mzML, MGF, native Thermo .raw and Bruker timsTOF .d. One of 17 bundled scoring models is picked from the file's activation, resolution and isobaric label. How it works

• Leaves the FDR to Percolator

  andes computes no production FDR. It writes the PIN, and Percolator, run by you or by --rescore, decides what passes q ≤ 0.01. Quick start

• Optional modes

  Co-isolated peptides (--chimeric), secondary modifications (--refine), and intact N-glycopeptides (--glyco). Modes

• Quantification built in

  TMT/iTRAQ reporter ions (--tmt) and label-free MS1 features with match between runs (--lfq), from the same run as the search. Quantification

The pipeline at a glance

flowchart LR
    S["Spectra<br/>mzML · .raw · .d"] --> M["Pick model<br/>per file"]
    F["FASTA"] --> I["Out-of-core<br/>fragment index"]
    M --> R["Retrieve and score<br/>top candidates"]
    I --> R
    R --> C["Optional passes<br/>chimeric · refine · glyco"]
    C --> P["PIN"] --> PC["Percolator<br/>owns the FDR"]
    PC --> O["PSMs · proteins · sites<br/>quantities (QPX)"]

Features built around trust in a result

• Knockoff variant FDR

  Variant claims get a null that makes the same mistakes as the claim: 65% to 2.5% true FDP on a genotyped dataset. Read more

• Chimeric cascade

  A second peptide from the residual spectrum: 43k to 66k Astral PSMs at an equal true FDP of 1%. Read more

• Error-controlled match between runs

  Every transferred quantity says what supports it and carries its own q-value. Read more

• TMT interference unmixing

  Per-scan co-isolation estimated from MS1, with no reference channel. Read more

All distinctive features The quantms ecosystem

Identifications, measured the same way

PSMs at Percolator q ≤ 0.01 on public reference datasets, rescored identically for every engine on the same 8-core machine. The conditions, dates and commits behind each bar are on the benchmarks page.

CPU time and memory

Same machine, same rescoring, 8 threads. Wall time, CPU time and peak memory for andes and Comet on four datasets; phospho is the one set where andes uses more memory.

Details and conditions: benchmarks.

Search and quantification

andes quantifies as it searches. Label-free runs get MS1 features and match between runs where every transferred value carries its evidence type and its own q-value; TMT runs get impurity-corrected reporter ions with a precision gate.

On the Astral module another tool has the lower median error; the quantification page shows both modules with their conditions.

Glycoproteomics

Intact N-glycopeptide search (--glyco) is benchmarked against five published tools on PXD065680, with entrapment error estimates for andes.

These are counts, not error rates; andes's entrapment FDP is 1.04–1.50% at a nominal 1%. See the glyco benchmark for the overlap, evidence tiers and every deviation from the paper's protocol.

Install

cargo install --git https://github.com/bigbio/andes --bin andes

Release archives for Linux, macOS and Windows are on the releases page. Details: Install.