Skip to content

fasta_lake.multi_omics.reproduce#

fasta_lake.multi_omics.reproduce #

Reproduce the published FL_MO headline numbers from committed artifacts.

The end-to-end FL_MO arm (per-sample SAGE searches over 306 samples) is too heavy to rerun here, so its final per-sample depth was frozen into committed per-sample TSVs by manuscript/verification/scripts/fdr_gate_benchmark.py:

manuscript/verification/fdr_gate_benchmark_FL_MO_razor_persample.tsv
manuscript/verification/fdr_gate_benchmark_FL_razor_persample.tsv

The published razor depth numbers (Methods §M5.5 / Results, defect-ledger D3) are the per-sample MEDIAN of the corrected-gate (protein_q <= 0.01) column:

FL_MO razor : 13,776 protein groups   (this module reproduces this)
FL    razor : 12,275 protein groups   (de-novo-only counterpart)

This module recomputes that median from the committed TSV so the headline number is regenerable from committed code, and asserts it matches 13,776.

Run: python -m fasta_lake.multi_omics reproduce or: python -m fasta_lake.multi_omics.reproduce

per_sample_median #

per_sample_median(tsv_path, column: str = DEPTH_COLUMN) -> tuple[float, int]

Return (median, n_samples) of an integer depth column in a per-sample TSV.

Source code in fasta_lake/multi_omics/reproduce.py 
48
49
50
51
52
53
def per_sample_median(tsv_path, column: str = DEPTH_COLUMN) -> tuple[float, int]:
    """Return (median, n_samples) of an integer depth column in a per-sample TSV."""
    with open(tsv_path) as f:
        rows = list(csv.DictReader(f, delimiter="\t"))
    vals = [int(r[column]) for r in rows]
    return statistics.median(vals), len(vals)

reproduce #

reproduce(verbose: bool = True) -> dict

Recompute and check the published FL_MO razor median.

Source code in fasta_lake/multi_omics/reproduce.py 
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
def reproduce(verbose: bool = True) -> dict:
    """Recompute and check the published FL_MO razor median."""
    fl_mo_median, n_mo = per_sample_median(FL_MO_RAZOR_TSV)
    fl_median, n_fl = per_sample_median(FL_RAZOR_TSV)
    lift_pct = 100.0 * (fl_mo_median - fl_median) / fl_median if fl_median else 0.0
    match = round(fl_mo_median) == PUBLISHED_FL_MO_RAZOR_MEDIAN

    result = {
        "fl_mo_razor_median": fl_mo_median,
        "fl_razor_median": fl_median,
        "n_samples_fl_mo": n_mo,
        "n_samples_fl": n_fl,
        "lift_pct": lift_pct,
        "published_fl_mo_razor_median": PUBLISHED_FL_MO_RAZOR_MEDIAN,
        "matches_published": match,
    }
    if verbose:
        print("FL_MO reproduction (gate: protein_q <= 0.01, razor):")
        print(f"  FL    razor per-sample median = {fl_median:.0f}  (n={n_fl})")
        print(f"  FL_MO razor per-sample median = {fl_mo_median:.0f}  (n={n_mo})")
        print(f"  per-sample lift               = +{lift_pct:.1f}%")
        print(f"  published FL_MO razor median  = {PUBLISHED_FL_MO_RAZOR_MEDIAN}")
        print(f"  REPRODUCED: {'MATCH' if match else 'MISMATCH'}")
    return result