Quantify a study

FastaLake assigns accepted peptides to a fixed set of study-level protein groups, then estimates a quantity for each group in each acquisition.

Method	Use it for	What changes
Sum — default	Direct summaries of accepted positive LFQ intensity	Adds assigned feature intensities; no sample normalization
directLFQ	Quantification from shared ion profiles across acquisitions	Normalizes and aligns ions; some quantities may be lost

Both methods use the same reporting dictionary. Missing output cells stay blank.

flowchart TD
    A["Accepted peptides across the study"] --> B["Fixed study dictionary"]
    B --> C["Assign each accepted LFQ feature once"]
    C --> S["Sum intensities"]
    C --> D["Normalize and align ions with directLFQ"]
    S --> Q["Study matrix and coverage report"]
    D --> Q
    Q --> V["QC and eligible PCA"]

Choose a method in the full workflow

The manifest runner uses sum by default. Add --quantification directlfq to select directLFQ. The selected matrix is study_groups/quantification/study_group_matrix.tsv; the original sum matrix remains in study_groups/.

Quantify existing searches

From the repository root in the installed environment:

fasta-lake group-study --search study_run/search --out study_groups \
  --quantification directlfq --threads 4

Or quantify an existing grouping bundle:

fasta-lake quantify-study --groups study_groups --method sum --out quantities_sum
fasta-lake quantify-study --groups study_groups --method directlfq --out quantities_lfq --threads 4

Each output directory must be new. The directLFQ route requires the optional backend; Docker includes it.

Open these outputs

File	What it tells you
study_group_matrix.tsv	Quantities for the declared acquisition roster
coverage.tsv	Groups quantified by each method and any coverage loss
summary.json	Method, settings and source bundle
qc/QC_overview.png	Coverage, missingness and PCA eligibility

For the full runner, QC lives under study_groups/qc/ and follows the selected method. The lower-level Python quantification function does not run QC itself.

The source grouping bundle retains peptide_evidence.tsv.gz and feature_assignments.tsv.gz. These distinguish accepted identifications from positive quantified features. An accepted peptide in the same acquisition gates LFQ inclusion; Sage's separate LFQ q_value is not an additional default filter.

Follow a quantity back to its evidence

A canonical peptide has one group within a fixed study dictionary. Modified peptides and charge states remain separate quantitative traces. Group labels do not establish a unique biological protein or species.

See quantification details for the ion merge keys, missingness, directLFQ support tables, backend settings and retained comparisons. Coverage loss deserves attention alongside precision; a lost quantity does not invalidate its peptide identification.