Writing Style & Claim Calibration (cc-writing-style)

When to trigger

• The Results read like a lab notebook rather than a mechanistic narrative
• The Discussion overstates therapeutic / clinical impact
• Claims outrun the evidence (single-system conclusions stated broadly)
• Nomenclature, tense, or section structure are inconsistent with Cell Press

Cell Press article structure (narrative roles)

Results: write the mechanism as a story

• Open each Results subsection with the question/claim, then the evidence, then the conclusion.
• Each paragraph maps to a figure/panel; every assertion cites a panel.
• Foreground causality and orthogonal validation ("loss of X reduced Y in cells and in PDX tumors") over a list of assays.
• Use past tense for what was done/found; present tense for established facts.

Claim calibration (the signature Cancer Cell discipline)

Match verb strength to evidence strength:

• Do not call a target "therapeutic" without in vivo efficacy and/or human evidence.
• Distinguish "necessary" from "sufficient"; do not conflate the two.
• State limitations explicitly (model caveats, single-cohort findings) — reviewers reward candor.
• Avoid "novel," "first," "unprecedented," "proves"; let the data carry the weight.

Nomenclature & house style

• Use approved gene/protein nomenclature (HGNC for human, MGI for mouse); italicize genes, roman for proteins; correct species capitalization (human BRCA1 vs mouse Brca1).
• Define each abbreviation once; keep terminology consistent across text, figures, and front matter.
• Cite using the Cell Press reference style (verify current format); reference key prior work fairly.
• Active voice and direct sentences; avoid hedging chains ("might possibly suggest").

Worked micro-edit: a Results paragraph in Cancer Cell voice

Before (assay catalog, verb outruns evidence):

To study MARK7, we performed RNA-seq, which showed many changes. Western blot confirmed knockdown. We also did a migration assay and saw differences. These novel results prove MARK7 is a promising therapeutic target that drives metastasis in patients.

After (claim → evidence → conclusion, calibrated):

MARK7 depletion reprogrammed the CAF secretome (Figure 3A; RNA-seq, n=4 donors), reducing lactate exporter transcripts, and this loss was confirmed at protein level with densitometry across three independent blots (Figure 3B). Functionally, MARK7 loss was required for CAF-driven tumor-cell migration in the co-culture system (Figure 3C). Because these data are cell-intrinsic, we tested the axis in vivo below before making any therapeutic statement.

The "after" ties each sentence to a panel and an n, uses "required for" for a single-system perturbation, and defers the translational verb until in vivo evidence exists — the Cancer Cell reflex of never letting the Results outrun the systems tested.

Discussion craft (where Cancer Cell papers over- or under-claim)

• Open the Discussion by restating the mechanism with direction, not by re-summarizing every figure.
• Devote one honest paragraph to model limitations: cell-line artifacts, xenograft immune context, cohort size and retrospective design, and whether efficacy is genetic (knockout) vs. pharmacologic.
• Separate what the data show from what a clinician would need — a preclinical vulnerability is a hypothesis for a trial, not a treatment recommendation.
• End on a proportionate significance line that a skeptical referee would sign off on.

Referee reflexes to preempt

Cancer Cell reviewers read prose against the figures. Common language flags they raise:

• "The authors write 'drives tumorigenesis' but show only correlation in the human cohort" — verb/evidence mismatch.
• "The Summary claims a therapy; the paper has no in vivo efficacy" — front-matter overreach.
• "Necessary and sufficient are used interchangeably" — mechanistic imprecision.
• "Every finding is 'novel' and 'striking'" — hype that erodes trust in the real result.
• "Mouse gene written as human (uppercase, non-italic)" — nomenclature sloppiness that signals weak rigor.

Fix these at the language layer before they become review comments.

Checklist

• Introduction is concise and ends on a clear hypothesis/gap
• Each Results paragraph maps to a figure and states claim → evidence → conclusion
• Causality and orthogonal validation are foregrounded
• Verbs match evidence level; no therapeutic claim without in vivo/human support
• "Necessary" vs "sufficient" used correctly
• Discussion states limitations and calibrated implications
• Gene/protein nomenclature and species formatting correct and consistent
• Hype words removed; tense usage consistent

Anti-patterns

• Results as an assay catalog with no narrative thread
• Overstated translational/therapeutic claims beyond the data
• Conclusions generalized from a single system
• "Novel/first/unprecedented/proves" sprinkled throughout
• Inconsistent gene nomenclature or species capitalization
• A Discussion with no limitations paragraph

Output format

【Results narrative】claim→evidence→conclusion per subsection? Y/N
【Claim calibration】overstated verbs flagged: [...]
【Therapeutic claims】backed by in vivo/human? Y/N
【Limitations】present in Discussion? Y/N
【Nomenclature】gene/protein/species correct & consistent? Y/N
【Hype words】removed? Y/N
【Next step】cc-cover-letter or cc-submission