[study] GEORGE2023_HEATWAVE_PLOIDY (PRJNA913164)

[sr320]

Claimed by aree start-study on Hete, 2026-10-11. Run and harmonization finished 2026-10-11. Blocked on sr320/AREE#24 (single-end / Tag-seq / design-table support). This branch contains that commit, so check-study-scope fails until #24 merges and this branch is rebased onto main.

Study

• study_id: GEORGE2023_HEATWAVE_PLOIDY
• Accession(s): BioProject PRJNA913164 (72 runs). The paper is George et al. 2023, Glob Chang Biol 29:6969, 10.1111/gcb.16880. The backlog said "no publication found"; the link rests on the BioProject description, not a full-text accession citation.
• Design: adult diploid and triploid ctenidium. Treatments: control (20 °C), single stressor (30 °C seawater) and multiple stressor (30 °C, then 6 h emersion at 44 °C). 12 per cell, single-end QuantSeq 3′ Tag-seq.
• Mode: raw_reanalysis, with --single_end --tag_seq
• Machine and storage used: Hete (16 GB, 8 cores), native -profile local,workstation, Nextflow 26.04.7. FASTQ, reference and work dir on /Volumes/Monarch (HFS+).

Result

Comparison	n (after exclusions)	padj < 0.05
diploid 30 °C vs 20 °C (meta_analysis_primary)	11 v 11	3,081 / 25,329
triploid 30 °C vs 20 °C	10 v 11	2,254 / 24,898
diploid 30 °C + 44 °C emersion vs 20 °C	10 v 11	1,688 / 25,058
triploid 30 °C + 44 °C emersion vs 20 °C	11 v 11	1,614 / 24,956

• Replicates are real: median within-group dispersion is 0.028–0.052 in all six groups. The pseudo-replicated IOCAS studies were at about 0.001.
• Agrees with the authors: 88–97% of each published DEG list changes in the same direction here, and 99.3–99.8% of genes significant in both analyses.
• Ploidies respond differently: fold changes correlate only r = 0.22–0.24 between diploids and triploids, consistent with the paper.
• Harmonized: 100,241 evidence records, 100% exact identifier mapping. In a thermal meta-analysis the primary is currently k = 1; LI2022_HEAT_PLOIDY (sr320/AREE#16) is the likely first pool partner.

Checklist

• aree validate-study passes (expected warnings: stress_response overrides the thermal_tolerance default)
• Sample sheet and FASTQ manifest generated by aree fetch-samplesheet, not hand-written. The design is joined from the authors' NCBI submission sheet, pinned to commit 24e7abd and committed; the BioSamples carry no treatment or ploidy.
• All 72 FASTQ MD5s verified against ENA before the run (18 needed re-download)
• Workflow manifests committed. Caveat: software_versions shows the container pins, not the native tools (Salmon 2.3.4, fastp 1.3.6). The real per-process versions are in process_versions.yml.
• Standardized result tables and HTML reports committed; no raw reads, work/ or results/
• qc_status: passed_with_warnings, analysis_status: harmonized
• Limitations and flags recorded: design comes from the authors' sheet; survival was measured per group, not per oyster (ambiguous_phenotype_definition); 3′ count scale; shared controls; single_timepoint
• Curation tests in tests/test_george2023_study.py (12; full suite 285 passed)
• No edits to shared code. Those are in Support single-end and 3' Tag-seq RNA-seq, and designs from authors' tables #24; rebase after it merges.

Curation decisions

• stress_response, not resilience. Mortality was measured (multiple stressor: 36.4% of triploids vs 14.8% of diploids died) but per ploidy × treatment group. The sequenced animals are survivors, so no comparison contrasts tolerant with susceptible animals.
• Within-ploidy contrasts, prespecified before any result. The primary is diploid 30 °C vs control: a single thermal stressor in the standard stock. The other three share controls and do not pool alongside it.
• Exclusions follow the authors. Their 8 outlier libraries (2_DESeq2_analysis-gigas-filtered.Rmd lines 111–116) are applied to every comparison, rather than an AREE-invented rule.
• Config fix. The first attempt's three non-primary runs were filed under the primary's id: params.rnaseq.comparison_id beats --comparison_id. They were discarded unharmonized, and the config now leaves that field null. A curation test now checks every manifest's recorded arms against the comparison it is filed under.
• Conflict of interest: sr320 is a co-author of the source paper.

🤖 Generated with Claude Code