Differential Hsp90-dependent gene expression is strain-specific and common among yeast strains.
A 0–100 reproducibility-quality score from the per-question grades, shown as a z-score: standard deviations above (+) or below (−) the mean of comparable assessments.
▸Reproduction agent’s raw note
PRELIMINARY (compute pending; «our HPC» VPN tunnel was down at start). Paper is well-described and HIGHLY reproducible: real data = SRA PRJNA606397 (BRIEF.md mined links GSE22269/lh3-bwa were false positives, corrected). Pipeline = Trimmomatic v0.36 -> Salmon (gcBias, R64-2-1 ORF transcriptome) -> DESeq2, thresholds FC>2 or <0.5 & padj<0.05. Design (C3) already CONFIRMED 1:1 from SRA metadata. Core targets: per-strain DEG counts (C2) and the headline 298-shared-genes number (C1), to be computed on «our HPC». Out of scope: western blots, TF one-hybrid (wet-lab), TF-motif enrichment + GO (secondary). Will update to reproduced/within-tol/mismatch after the «our HPC» run.
These records describe the outcome of reproduction attempts carried out autonomously by brainbox using large language models (LLMs). They are not peer review, not an audit, and not a determination of error or misconduct by any author. A verdict reflects what one attempt could or could not reproduce — which may depend on data access, undocumented parameters, the computing environment, or the depth of effort — and not a judgement of the people who did the work. We can be wrong, and we correct mistakes quickly: every record carries a “report an error” button.
✎ I am an author of this paper
Updated or fixed a deposit, or is there an erratum? Ask us to re-run the metrics. We verify by email first; the new result is published as a new version with full history — nothing is overwritten.
Provenance — full disclosure
When this reproduction was carried out, which methodology version was used, and by whom — so the record can be audited and checked independently.
- Reproduced
- 2026-06-19
- Rubric version
- not recorded
- Assessed by
- —
- Last updated
- 2026-07-29
Provisional, curator- or AI-assessed, and independently checkable. A reproduction outcome states what one attempt could reproduce — not a judgement of the authors.
Deep full-text extraction
Model: opusThe authors test how common Hsp90-dependent differential gene expression is in natural yeast populations, hypothesizing that Hsp90's buffering and potentiating effects on genetic variation are revealed as strain-specific transcriptional differences when Hsp90 is compromised.
- ★ Hundreds to over a thousand genes show Hsp90-dependent, strain-specific differential expression across five diverse yeast strains. finding
- ★ Both Hsp90-buffering and Hsp90-potentiating events occur in natural yeast populations. finding
- ★ Specific transcription factors (e.g., Msn2/Msn4, Com2, Rap1, Dot6, Tod6) mediate strain-specific Hsp90-dependent gene expression, with Hsp90-interacting TFs significantly enriched among candidates. mechanism
- ★ Hsp90 inhibition leads to strain-specific upregulation of Msn4 target genes (higher in SK/NA than ML). finding
- ★ Hsp90-dependent differential expression is correlated with phenotypic change and manifests under environmental stress. finding
- A bioinformatics genome-wide promoter-scan pipeline predicts strain-conserved TF binding sites, validated against ChIP-seq data. method
- 298 Hsp90-dependent differentially expressed genes are shared by at least four strains, representing general/ancestral Hsp90 responses. finding
- The transcriptome divergence under Hsp90 inhibition departs from the genetic distance tree, consistent with the capacitor hypothesis. finding
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| bulk RNA-seq | five diverged S. cerevisiae strains (S288C/LAB, SK, ML, NA, WA) | Geldanamycin (GdA) Hsp90 inhibitor, 50 μM | genome-wide mRNA expression / fold-change (GdA+/GdA−) | — |
| bulk RNA-seq | ML strain of S. cerevisiae | Geldanamycin, 25 μM (to avoid slow-growth confounder) | transcriptome profile compared across strains/conditions | — |
| v-Src kinase activity assay | five yeast strains | Geldanamycin ± treatment | Hsp90 client (v-Src) kinase activity as proxy for Hsp90 activity | — |
| Western blot | five yeast strains | normal growth conditions | Hsp90 protein abundance | — |
| growth assay | ML and other yeast strains | Geldanamycin 50 μM vs 25 μM | cell growth rate | — |
| computational TFBS prediction (promoter motif scan) | genomes of five yeast strains | none | predicted strain-conserved transcription factor binding sites | — |
| ChIP-seq (external validation data) | yeast (laboratory strain) | none | overlap with predicted TFBS candidates | — |
- – Hundreds to more than one thousand genes changed expression on GdA treatment across strains; LAB strain had the fewest. FC >2 or <0.5, adj pvalue <0.05
- – 25 μM GdA-treated ML transcriptome more similar to 50 μM GdA-treated ML than to other strains. ρ = 0.83 (ML) vs 0.81/0.74/0.72/0.59 (NA/WA/SK/LAB)
- ▲ Hsp90-interacting TFs significantly enriched among candidate TFs. 7/21 vs 23/187, p=0.0011
- ▲ Msn2/Msn4 target genes significantly upregulated in SK and NA relative to ML on Hsp90 inhibition.
- – Expression tree in normal condition highly similar to genetic distance tree; under Hsp90 inhibition it diverges. ρ = 0.71 (−GdA) vs ρ = −0.11 (+GdA)
- – 298 Hsp90-dependent DE genes shared by ≥4 strains with seven enriched GO biological processes. 298 genes; 7 GO terms
- – Predicted TFBS validated by ChIP-seq overlap. ~58% overlap; 96% conserved across strains
- ▲ GdA-treated SK cells had the highest fraction of up-regulated stress-response genes, but general stress response not triggered. 27/133 = 20%
- correlation rho >0.98, pvalue <2.2e-16 (Spearman correlation between three biological replicates per sample)
- correlation ρ = 0.83, 0.81, 0.74, 0.72, 0.59 (25 μM GdA ML vs 50 μM GdA ML/NA/WA/SK/LAB)
- pvalue p = 0.0011 (hypergeometric test, Hsp90-interacting TF enrichment (7/21 vs 23/187))
- correlation ρ = 0.71 (expression tree (−GdA) vs genetic distance tree)
- correlation ρ = −0.11 (expression tree (+GdA) vs genetic distance tree)
- count 298 genes (Hsp90-dependent DE genes shared by ≥4 strains)
- count 27/133 = 20% (stress-response genes up-regulated in GdA-treated SK cells)
- other 50,000–85,000 SNPs (pairwise genetic divergence among the five strains)
Statistical methods review
Model: sonnetA neutral, descriptive read of the statistical approach — what was done, and (for shared learning, not as criticism) what could also have been done.
The paper compared transcriptomes (RNA-seq, three biological replicates per strain/condition) of five diverged yeast strains with and without the Hsp90 inhibitor Geldanamycin (GdA), calling differentially expressed genes by a fold-change threshold (>2 or <0.5) combined with a multiple-testing-corrected p-value (<0.05). Strain-specific and strain-pair-specific effects were further examined using correlation analyses, GO enrichment testing, a hypergeometric test for transcription-factor enrichment, and two-sided Wilcoxon rank-sum tests (with Bonferroni correction) comparing target-gene expression distributions between strains. Results are reported largely as significance thresholds and fold-changes, with boxplots (median, IQR) used for some comparisons.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| Fold-change threshold with multiple-testing-corrected p-value (test/model not named in text) | Differentially expressed genes on GdA treatment vs untreated, each strain (Figure 1A, S1B, S1C, S3, Table S5) | three biological replicates per sample | not stated |
| Spearman's correlation | Reproducibility between biological replicates; comparison of expression distance trees to genetic distance trees (Table S1; Figure S1) | three biological replicates | not stated |
| GO enrichment analysis with multiple-testing correction | Enrichment of biological processes among Hsp90-dependent and strain-pair-specific genes (Figure 1B, 1C, Table S6) | — | not stated |
| Hypergeometric test | Enrichment of Hsp90-interacting TFs among candidate TF list (p = 0.0011) | 7/21 candidates vs 23/187 genome-wide | not stated |
| Two-sided Wilcoxon rank-sum test (Mann-Whitney U), Bonferroni-adjusted | Differences in Hsp90-dependent expression of TF target genes between strains (Figure 2, 3A, Tables S11, S12) | — | not stated |
-
Differential expression between GdA-treated and untreated samples was called using a fold-change threshold (>2 or <0.5) plus a multiple-testing-corrected p-value, without naming the underlying statistical/count model.↳ Could also: A model-based RNA-seq tool such as DESeq2 or edgeR (negative binomial generalized linear model with variance/fold-change shrinkage) — These tools explicitly model count-based mean-variance relationships and provide moderated fold-change estimates, which can be particularly informative when replicate numbers are modest.
-
Strain-pair-specific (buffered/potentiated) genes were classified using a ratio-of-fold-changes threshold (>2 or <0.5) rather than a formal statistical interaction test.↳ Could also: A statistical interaction test within a generalized linear model (e.g., a likelihood-ratio test on a strain x treatment interaction term) — This would let the strength of the strain-specific Hsp90-dependence be assigned a p-value directly, complementing the fold-change-ratio heuristic used to flag candidate genes.
-
Differences in Hsp90-dependent expression of TF target genes across strains were tested pairwise with a two-sided Wilcoxon rank-sum test.↳ Could also: A linear model or mixed-effects framework (e.g., limma, or ANOVA with strain and treatment as factors) fit across all strains simultaneously — A single joint model can estimate strain and treatment effects (and their interaction) together, which can be a useful complement to multiple pairwise nonparametric comparisons.
-
Multiple-testing correction for DE-gene calls and GO enrichment was described generically as 'multiple testing correction adjusted p-value' without naming the specific method.↳ Could also: Explicitly reporting the method, e.g., Benjamini-Hochberg FDR versus Bonferroni — Naming the method clarifies how conservative the adjustment is; BH-FDR is commonly used for genome-wide transcriptomic screens and can be more powerful than Bonferroni while still controlling for multiple comparisons.
-
RNA-seq comparisons were based on three biological replicates per strain/condition.↳ Could also: Reporting a power calculation or including additional replicates for key comparisons — This can help readers gauge the precision of fold-change estimates, complementing the correlation-based reproducibility checks already reported.
-
Results are reported largely through significance thresholds and fold-change cutoffs rather than continuous effect-size estimates with uncertainty intervals.↳ Could also: Reporting log2 fold-changes together with confidence intervals — This would convey the precision of each estimate in addition to whether it crosses a significance threshold.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-37138775
Paper: Hung et al. 2023, iScience. Differential Hsp90-dependent gene expression is strain-specific and common among yeast strains. PMID 37138775 / PMC10149407 / DOI 10.1016/j.isci.2023.106635
Correction to BRIEF.md mined links (important)
The brief's auto-mined Code: lh3/bwa and Data: GSE22269 are false positives:
GSE22269= a 2010 SMER / rapamycin yeast microarray study (PMID 20581845), unrelated.lh3/bwa= generic DNA short-read aligner; not this paper's code, wrong tool category (this is an RNA-seq quantification + DESeq2 study, no genome alignment of reads to a reference with BWA is described).
Real artifacts (from PMC Resource Availability + STAR Methods):
- Data: SRA PRJNA606397 (SRP249481), runs SRR11091801–SRR11091831.
- Authors' code: github.com/phhung1989/motif-scan (TF-motif scanner — secondary).
- RNA-seq DE pipeline: standard third-party tools (Trimmomatic + Salmon + DESeq2) — a P16-valid reproduction (existing tool on the paper's own data).
Dataset
PRJNA606397: 31 Illumina NextSeq 500 runs. 30 RNA-seq (single-end 75 bp) + 1 DNA-seq
(SRR11091807, WA genomic, paired-end — excluded from DE). Design = 5 S. cerevisiae strains
(SK/Y12, ML/UWOPS05-217.3, LAB/R1158, WA/DBVPG6044, NA/YPS606) × 2 conditions
(GdA = geldanamycin Hsp90 inhibitor; control = DMSO in CSM medium) × 3 biological replicates.
See data/samples.tsv for the full run→strain→condition map.
IN SCOPE (pipeline-derived, will attempt)
| id | result | pipeline |
|---|---|---|
| C1 | 298 genes shared by ≥4 strains among Hsp90-dependent DEGs (headline number) | Trimmomatic→Salmon→DESeq2 per strain, then intersect DEG sets across strains |
| C2 | Per-strain DEG counts "hundreds to >1000" on GdA treatment (FC>2 or <0.5, padj<0.05) | DESeq2 GdA vs control per strain |
| C3 | Experimental design: 5 strains × 2 conditions × 3 replicates (30 RNA-seq libraries) | metadata verification (already confirmed from SRA) |
Thresholds (from Methods): |fold change| → FC > 2 or < 0.5; multiple-testing adjusted p-value < 0.05. Reference: orf_coding_all.fasta R64-2-1. Salmon in gcBias mode.
OUT OF SCOPE (not attempted — wet-lab / manual / external)
- Western blots (Hsp90/Hsp82 protein levels) — wet-lab.
- Transcription-factor one-hybrid assays — wet-lab.
- TF-motif enrichment (21 TFs / 11 TFs) via motif-scan + PWM scanning — secondary analysis layered on top of the DE result; depends on promoter extraction + curated PWM cutoffs; may attempt as a stretch goal only after the core DE result is reproduced.
- GO enrichment of the 298 common genes (Table S6) — downstream annotation, low priority.
- DNA-seq run SRR11091807 (WA genome) — not part of the expression result.
- GSE147927 ChIP-seq — cited external reference, not reprocessed.
Reproduction strategy
- Download 30 RNA-seq FASTQ + R64-2-1 transcriptome on «host» → «infra».
- Build conda env (salmon, trimmomatic, fastqc) on front1.
- SLURM: Trimmomatic (SE) → Salmon quant (gcBias, validateMappings) per sample.
- R/DESeq2: import via tximport, per-strain GdA-vs-control, apply thresholds, count DEGs.
- Intersect DEG sets → genes DE in ≥4 strains → compare to 298.
- If Table S5 (DEG lists) obtainable, compute overlap/Jaccard against the actual gene sets.
Assessments & scoring basis
Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.
Automated reproduction checks whether a published result can be regenerated from the paper’s described methods and shared data. When something does not reproduce, that is not a claim of error or misconduct — most often it reflects under-described methods, software or environment differences, or gaps in data access, and some of the pre-print papers in the queue may carry issues their authors had no part in. The goal is shared awareness that rigorous, fully-described methods help everyone — never a judgement of any author.
Are you an author? We would genuinely like to hear from you — to clarify the record, add data or code, re-run the pipeline after an accession update, and publish your response right next to the assessment. Everything here is open and auditable.
🚩 Report an error in this record
Spotted something wrong — a verdict you’d contest, a data or value error, or a private detail that slipped through? Tell us, with a short justification. Authors and readers are equally welcome to write in; we review every report.
Prefer email, or the form below not working? Contact us at support@doesitreproduce.com.
Reproduction footprint
claude-opus-4-8Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.