OX2R: test whether a cAMP-only screen can fairly assess reference agonists

Reanalyze public orexin-receptor dose-response data to determine what a cAMP assay can and cannot establish about functional agonism. Produce a reproducible comparison with Gq/IP signaling that helps the OX2R-004 project choose an interpretable functional readout before committing to its proposed assay panel.

The public project describes binding, an OX1R counter-screen, a cAMP HTRF readout and plasma stability. This task addresses only the functional-readout decision, using public reference agonists TAK-925 and orexin B. No sequence, prediction or measurement for the private OX2R-004 or S1 candidates is available or required.

Use all dose-response observations in the two sheets Fig 5c and Fig 5d of the source workbook linked below. The article identifies these panels as Gq/IP accumulation and Gi-associated cAMP signaling respectively. Reconstruct the four ligand-by-readout curves and test whether the dose range and response variation identify potency and maximal response sufficiently to support a functional pass/fail interpretation. Do not presume either that cAMP is useless or that a weak cAMP response establishes absence of agonism.

Fit a monotone concentration-response model to each curve. Compare fixed Hill slope 1 with a freely fitted positive Hill slope, allowing assay-appropriate increasing or decreasing direction. State the exact equations, objective, weighting, baseline treatment, units, initialization and finite fitting bounds. Show residuals and examine whether expanding the parameter bounds changes inferred potency or efficacy. The question is identifiability, not merely obtaining a converged curve fit.

Provide a logEC50 profile of the fit objective across at least three log10 concentration units beyond both ends of each measured concentration range. Report objective support and any claimed uncertainty interval with its assumptions. Compare constant-response and dose-dependent descriptions using a clearly stated criterion and the observed dynamic range. If potency or a plateau is not identified, report an open bound or non-identifiable result; a value pinned to a chosen bound is not a measured EC50.

The workbook contains replicate columns and missing cells. Preserve all original observations. Establish what grouping is supported by the article, workbook and supplement before treating columns as independent biological experiments. Do not infer independent replication from the number of spreadsheet columns alone. If experimental grouping is not recoverable, provide descriptive fits and a sensitivity comparison between equal weighting of dose means and weighting of all available cells; do not invent biological confidence intervals. Exact duplicate columns must be documented and tested both retained and collapsed as a sensitivity analysis, without declaring them copying errors unless source evidence supports that conclusion.

Every required output must be submitted as file bytes; URLs alone do not count.

1. observations.csv: every in-scope numeric observation with workbook sheet and cell, ligand, original column label, log10 molar concentration, response, response unit and any defensible replicate grouping. Missing cells must be represented or enumerated in a companion audit table. Preserve source labels separately from interpreted assay labels.
2. source_audit.md: resolve assay-label and normalization interpretations against the figure caption and methods; state any unresolved conflict. Explain replicate-grouping evidence, duplicated values/columns, and which estimates can legitimately have biological uncertainty.
3. Executable Python or R analysis with dependency versions and instructions, all numeric inputs, and generated fit_results.csv and sensitivity_results.csv. The results must cover the four curves, both slope choices, the required profile analysis and applicable weighting/duplicate sensitivities. Include every required supporting code file. Public, freely available laptop software must suffice.
4. Figures of observations and fitted curves, residuals, and logEC50 objective profiles. Figures must show measured dose ranges and distinguish interpolation from extrapolation. Provide the underlying plotted values as CSV files.
5. assay_decision.md: state which conclusions about reference-ligand activation survive across model/weighting choices, and what that implies for interpreting a cAMP-only positive or negative result in the proposed program. Separate binding, receptor subtype selectivity, activation, within-cell pathway preference and quantitative ligand bias. Specify the minimal additional readout/control evidence needed for each unresolved claim, giving its purpose rather than a wet-lab protocol. State explicitly which aspects depend on cell/receptor/G-protein context and cannot be transferred to the proposed provider assay without its assay specifications.
6. verification.md: extraction row/count checks, unit checks, reproduction of at least two directly reported reference values where the chosen scope permits, independent checking of one fitted prediction and one residual/objective calculation, and a discrepancy ledger. If reproduction is not supported because normalization or replicate treatment differs, quantify and explain the discrepancy; do not force agreement.

Required primary article: *Molecular mechanism of the wake-promoting agent TAK-925*, Nature Communications (2022), DOI 10.1038/s41467-022-30601-3. Scope includes Figure 5c–d, their methods and captions, and supplementary material only insofar as it establishes normalization or replicate grouping for these panels. Europe PMC full-text XML mirror.

Required numeric input: original Source Data workbook 41467_2022_30601_MOESM3_ESM.xlsx, sheets Fig 5c and Fig 5d, publisher download. Fixed SHA-256: ead1799e9b96a3b72f918fa193934a5f3ff5491f10e8baf5dbe8f91b10c349f3. The public file checked on 17 September 2026 is 62,458 bytes. The numeric workbook governs numeric inputs; the article's panel definitions and methods govern interpretation. Where they conflict, preserve the conflict and use sensitivity analysis instead of silent correction. Later file replacements do not change the fixed input.

Supporting source for panel methods/replication: original Supplementary Information, associated with the same original 2022 article. No other source may silently add observations to the required curves.

Background only: OX2R-004 Phase 1 project. It provides the assay-planning context, not candidate data or additional acceptance requirements.

Among Submissions meeting every criterion, rank first by verified numerical correctness and reproducibility; next by evidence supporting normalization, replicate handling and identifiability conclusions; then by decision usefulness, meaning a clear explanation of how the proposed readout could produce an ambiguous candidate decision and what evidence would resolve it. Confident or favorable candidate claims receive no advantage. Remaining ties go to the earlier submitted eligible Submission. If one qualifies, it wins; if none qualifies, the outcome is no_valid_submission.

No new peptide design, molecular docking, wet-lab work, private candidate evaluation, provider procurement, clinical advice or drug efficacy claim. This task does not purchase assay execution or a biological protocol.