Review papers with raw data transparency

Paper review summary

This critique examines strengths, weaknesses, reproducibility, and biological interpretation of Maria Abran et al., Mutational Landscape of TP53 Across Cancer Types (preprint DOI 10.1101/2025.08.12.669884). The dataset is TCGA PanCancer via cBioPortal (downloaded 9 Aug 2025): 4,245 TP53 mutation events in 10,967 tumor samples across 27 cancer types; missense mutations dominate (64.3%) with cancer-type specific patterns (e.g., CNS highest missense percent 76.5%; female reproductive cancers highest burden n=1,039) as reported by the authorsAbran et al mutational summary and methods

Quick critical takeaway: the paper provides a useful descriptive atlas of TP53 mutation counts and proportions across TCGA cohorts but stops short of deeper functional, clinical, or co-mutation context; methods rely on Excel and simple aggregation which limits reproducibility and downstream analyses (see long review for details).

Detailed review and critique Mutational Landscape of TP53 Across Cancer Types

1 Authors claims and primary data

• The study retrieved TP53 somatic mutation data from cBioPortal (TCGA PanCancer Atlas) downloaded 9 August 2025 and analyzed 4,245 TP53 mutation events in 10,967 unique tumor samples across 32 studies covering 27 cancer types; mutations were classified into six categories and cancer types grouped into ten broader categoriesMethods and data retrieval from Abran et al
• Key reported numbers: overall missense 2731 64.3% nonsense 546 12.9% frameshift 522 12.3% splice-site 335 7.9% in-frame 91 2.1% fusion 20 0.5% ; female reproductive cancers showed largest mutation burden n=1039 and CNS cancers had highest percent missense 76.5% (group-level breakdown reported in Results) Results and mutation counts from Abran et al

2 Strengths

• Large aggregated sample from TCGA PanCancer Atlas via cBioPortal increases breadth of cross–cancer comparisons and aligns with prior knowledge that TP53 is the most commonly mutated tumor suppressor across many epithelial cancersAuthors alignment with TCGA ICGC patterns
• The paper provides clear descriptive breakdowns by mutation class and by cancer group that can serve as a reference table for subsequent functional or clinical correlation studiesReference tables and figures in Abran et al

3 Major weaknesses and methodological limitations

1. Reproducibility and provenance — data were downloaded from cBioPortal but authors do not provide the exact TSV file, query filters, study identifiers list, or a script/notebook; preprocessing was done in Microsoft Excel, which is not a robust reproducible pipeline for genomic data (risk of silent errors, sorting/formatting issues, loss of metadata). This lowers reproducibility and auditabilityData availability and preprocessing limitations
2. Limited analytical depth — analyses are descriptive frequency counts and proportions; the paper explicitly notes it did not integrate clinical outcomes, co-mutation patterns, mutational signatures, variant functional annotation (hotspot vs truncating vs dominant negative/gain-of-function), or pathway enrichment; absence of these reduces biological and clinical interpretabilityAuthors note on missing integrative analyses
3. Grouping strategy and sample heterogeneity — collapsing 27 distinct cancer histologies into ten broad groups simplifies visualization but risks obscuring tumor-specific signals (e.g., serous endometrial cancer vs endometrioid have very different TP53 spectra) and may misrepresent cancers with low sample counts; authors do not present group-level sample size confidence intervals or per-study breakdowns to expose heterogeneityGrouping explanation and granularity in Abran et al
4. No statistical testing or uncertainty quantification — results report counts and percentages but lack confidence intervals hypothesis tests or correction for differing sample sizes; this makes it unclear whether observed differences (eg CNS 76.5% missense vs head and neck 57.8%) are statistically significant or driven by sampling variabilityLack of formal statistics in Abran et al
5. Functional annotation missing — missense mutations dominate but missense is a heterogeneous class spanning benign passenger substitutions hotspot DNA-binding domain mutants that confer dominant negative or gain-of-function effects; without mapping to protein coordinates domain annotations or functional assays the biological meaning of missense dominance is incomplete (paper acknowledges lack of functional consequence exploration)Missing functional variant annotation in Abran et al

4 Specific methodological and interpretational suggestions

1. Replace Excel preprocessing with a public, versioned script (R Python) that documents exact cBioPortal query parameters (study IDs sample filters mutation filters), includes data checks, and writes processed TSV/CSV to a public repository (eg Zenodo/GitHub) to achieve reproducibility and FAIR standardsReproducibility note from Abran et al
2. Annotate all variants with genomic coordinates protein positions (Hg38), map to TP53 functional domains and known hotspots (eg R175 R248 R273) and assign predicted consequence classes (truncating vs nontruncating hotspot) using established tools (VEP Annovar HotSpot3D) to distinguish passenger from likely driver events (this would change biological conclusions about missense prevalence) — existing literature demonstrates clinical differences by TP53 variant class and hotspot status (see e.g. endometrial cancer TP53 variant spectrum differences) Authors call for functional annotation integration
3. Perform per-cancer statistical comparisons with confidence intervals and tests (eg chi-square Fisher exact for categorical proportions with multiple testing correction) and present per-study heterogeneity (forest plots) to ensure group differences are robust to sampling and study-to-study variability.
4. Integrate co-mutation analysis, tumor mutation burden, mutational signatures (SBS), and clinical endpoints (survival treatment response) when available; TP53 impact is context-dependent and co-drivers and TMB/immune landscape modify its clinical effect.

5 Reproducibility assessment and score

I estimate reproducibility score 6/10 because raw source (cBioPortal) is a public resource and download date is provided, but absence of scripts raw TSVs and use of Excel reduce reproducibility and increase risk of transcription/sorting errors; a programmatic pipeline and data deposition would raise this to 9–10/10Reproducibility notes in Abran et al

6 Biological interpretation and caveats

• Missense dominance is expected for TP53 given many recurrent DNA-binding domain hotspots that change transcriptional programs; however the paper does not distinguish hotspot gain-of-function or dominant negative mutants from likely passenger missense mutations — this distinction matters clinically because specific TP53 mutants confer distinct prognoses and therapy responses (eg some hotspots associate with worse outcomes) — without this annotation the statement that missense mutants dominate is true but biologically incompleteMissense finding and limits from Abran et al
• Group-level observations such as high TP53 burden in female reproductive and gastrointestinal cancers are consistent with prior TCGA/ICGC observations but may reflect histology composition (eg high serous ovarian and copy-number-high endometrial subtypes are TP53-enriched) rather than a tissue-intrinsic propensity — authors should disambiguate histology-specific subtypes within groupsCaveat about histology composition from Abran et al

7 What would falsify the main conclusions

The paper's primary descriptive claim (missense mutations predominate and cancer-type–specific burdens exist in TCGA PanCancer data) would be falsified if an independent curated reanalysis of the identical TCGA cohorts showed a substantially different distribution (eg missense <50%) or if per-study heterogeneity explained away group differences; such falsification is feasible with a reproducible pipeline and should be attempted prior to clinical translationAuthors falsification statement

8 Practical next steps and high value followups

1. Recompute variant-level annotations programmatically mapping to HG38 protein coordinates add domain/hotspot labeling, ClinVar/OncoKB/CBioPortal hotspot flags and categorize into truncating hotspot nonhotspot missense; test enrichment per cancer-type.
2. Perform co-mutation clustering and compute mutual exclusivity/cooccurrence with major oncogenic drivers (eg KRAS EGFR PIK3CA PTEN) and tumour mutation burden correlations to contextualize TP53 role in each cancer group.
3. Integrate outcome data where available to test prognostic value of TP53 variant class (truncating vs hotspot missense) with multivariable Cox models adjusted for stage/age/treatment.

9 Visual reproduction of key descriptive figures (interactive)

10 Overall evaluative scores

11 Short conclusion and confidence

Conclusion: The manuscript provides a clear descriptive inventory of TP53 mutation counts and proportions across TCGA PanCancer cohorts and is a useful starting reference, but it is limited by lack of reproducible programmatic pipelines, absence of variant-level functional annotation, and missing statistical inference and clinical integration; these gaps reduce mechanistic and translational value until addressed. Confidence in the descriptive numbers (as reported by authors) is moderate because the source is public TCGA data but reproducibility/reporting gaps leave room for data-processing errorsAuthors final discussion and limitations