diagFDR report

diagfdr (Python) {{ version }} · {{ date }} · headline threshold α = {{ params.alpha_main }}

Executive summary

Alerts motivate inspection; they are not pass/fail verdicts. Items 1–4 of the reporting standard are internal checks (specification and consistency with the exported decoys); they cannot show that the decoys behave like incorrect targets — only external evidence (entrapment, known composition) can.

{{ row.list }} — {{ row.status }}: {{ row.interpretation }}

Lists

{% for nm, m in lists.items() %} {% endfor %}

List	Unit	Scope	q-value source	Export ceiling	Two-pass	n rows	n decoys	Provenance
{{ nm }}	{{ m.unit }}	{{ m.scope }}	`{{ m.q_source }}`	{{ m.q_max_export if m.q_max_export is not none else "unknown" }}	{{ m.two_pass if m.two_pass is not none else "unknown" }}	{{ "{:,}".format(m.n) }}	{{ "{:,}".format(m.n_decoy) }}	{{ m.provenance }}

Headline summary (Table)

{% for r in headline %} {% endfor %}

List	T_α	D_α	Decoy ratio D/T	D_α^-1/2	FDP-CV benchmark	D_α,win	IPE	π_D (low conf.)	Status
{{ r.list }}	{{ "{:,}".format(r.T_alpha) }}	{{ "{:,}".format(r.D_alpha) }}	{{ "%.3f%%"\|format(100*r.FDR_hat) if r.FDR_hat==r.FDR_hat else "n/a" }}	{{ "%.3f"\|format(r.support_index) }}	{{ "%.3f"\|format(r.CV_benchmark) }}	{{ r.D_alpha_win if r.D_alpha_win==r.D_alpha_win else "n/a" }}	{{ "%.4f"\|format(r.IPE) if r.IPE==r.IPE else "n/a" }}	{{ "%.3f"\|format(r.pi_D_hat) if r.pi_D_hat==r.pi_D_hat else "n/a" }} ({{ r.flag_equalchance }})	{{ r.status }}

Decoy ratio: a consistency check of the exported q-values on the reported list (expected ≤ α by construction), not the realised error. FDP-CV benchmark √2·D_α^-1/2: an independence reference for the dispersion of the realised FDP across experiments, neither a confidence interval nor a bound for this list. IPE: decoy-reference PEP concordance error (target mean PEP vs D_b/T_b per PEP bin), not a calibration against the truth. π_D: decoy fraction among entries with q in [{{ params.low_conf[0] }}, {{ params.low_conf[1] }}]; evaluable only for single-pass paired competition.

{% if sentences %}

Reporting sentences (five-item standard)

“{{ s }}”

{% endif %}

Alert basis

{% for r in basis %}{% endfor %}

Flag	Diagnostic	Threshold	Basis
{{ r.flag }}	{{ r.diagnostic }}	{{ r.threshold }}	{{ r.basis }}

Stability across thresholds

{% for r in stability %} {% endfor %}

List	α	T_α	D_α	D/T	FDP-CV benchmark	D_win	Elasticity J(α, {{ 1+params.eps }}α)
{{ r.list }}	{{ r.alpha }}	{{ "{:,}".format(r.T_alpha) }}	{{ "{:,}".format(r.D_alpha) }}	{{ "%.4f"\|format(r.FDR_hat) if r.FDR_hat==r.FDR_hat else "n/a" }}	{{ "%.3f"\|format(r.CV_benchmark) }}	{{ r.D_alpha_win if r.D_alpha_win==r.D_alpha_win else "n/a" }}	{{ "%.3f"\|format(r.jaccard) if r.jaccard==r.jaccard else "n/a" }}

{% if eq_bands %}

Equal chance by q-value band

{% for r in eq_bands %}{% endfor %}

List	q band	n	n decoy	Decoy fraction	mean q
{{ r.list }}	{{ r.qbin }}	{{ "{:,}".format(r.n) }}	{{ "{:,}".format(r.n_decoy) }}	{{ "%.3f"\|format(r.decoy_frac) }}	{{ "%.4f"\|format(r.q_mean) }}

{% endif %} {% if pep_bins %}

PEP concordance by bin (IPE)

{% for r in pep_bins %}{% endfor %}

List	PEP bin	T_b	D_b	mean target PEP	D_b/T_b	used
{{ r.list }}	{{ r.bin }}	{{ "{:,}".format(r.n_target) }}	{{ "{:,}".format(r.n_decoy) }}	{{ "%.4f"\|format(r.pep_mean_targets) if r.pep_mean_targets==r.pep_mean_targets else "n/a" }}	{{ "%.4f"\|format(r.e_b) }}	{{ r.used }}

{% endif %} {% if entrap %}

External validation (entrapment)

{% for r in entrap %}{% endfor %}

List	α	T_α	E_α	Raw FDP (lower bound)	Wilson 95%	r	Corrected (upper bound on average)
{{ r.list }}	{{ r.alpha }}	{{ "{:,}".format(r.T_alpha) }}	{{ r.E_alpha }}	{{ "%.3f%%"\|format(100*r.fdp_raw) }}	[{{ "%.3f"\|format(100r.lo) }}, {{ "%.3f"\|format(100r.hi) }}]%	{{ "%.3f"\|format(r.r) if r.r==r.r else "n/a" }}	{{ "%.3f%%"\|format(100*r.fdp_corrected) if r.fdp_corrected==r.fdp_corrected else "n/a" }}

Wilson intervals are descriptive and ignore clustering of errors within protein groups; the corrected value assumes the entrapment proteome is a valid model of incorrect matches and uses the size correction of Wen et al. (2025).

{% endif %}

Parameters

Reference: Chion M, Godmer A, Douché T, Matondo M, Giai Gianetto Q (2026). diagFDR: Assessing False Discovery Rate Claims in Proteomics with Scope, Calibration and Stability Diagnostics. bioRxiv 10.64898/2026.04.16.718468. R package: https://cran.r-project.org/package=diagFDR