RSK1 is an exploitable dependency in myeloproliferative neoplasms and secondary acute myeloid leukemia.
The main results reproduced, with only marginal, non-material deviations.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
- ✓Same input data as the authors
- ✓Reported values are derivable from the shared data
- ✓The central claim held under reproduction
- 🟡Reported values were only indirectly comparable
- 🟡A deviation arose in the data or preprocessing
- 🟡A deviation was attributed to the published material
- 🟡The deviation was non-trivial in magnitude
- 🟡Overall, the reproduction showed a material discrepancy
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
PARTIAL (core pipeline reproduced well). The scaffold accession GSE135902 is a TEXT-MINING FALSE POSITIVE (2019 CMML-monocyte RNA-seq PMID 31648319, used by this paper only as a Fig-6e reference cohort). The paper's REAL data is SuperSeries GSE228760; the one publicly-runnable named pipeline (ENCODE-DCC/atac-seq-pipeline, a third-party tool = valid per P16) applies to its ATAC subseries GSE229220 (THP-1 DMSO x3 vs PMD-026 5uM 24h x3). Reproduced the ENCODE ATAC algorithm end-to-end on the paper's own raw FASTQ on «our HPC»/«infra»: all 6 samples aligned 96.5-97.1% (bowtie2 GRCh38 --very-sensitive -X2000) and passed ENCODE filtering; pooled MACS2 (BAMPE, Tn5 shift) called 133092 (DMSO) / 123950 (PMD) peaks. RESULTS: C2 - the strongest anchor - our reproduced pooled fold-change signal correlates with the authors' OWN deposited fc_signal bigWigs at Pearson ~0.90 / Spearman ~0.91 genome-wide (305550 x 10kb bins): the named pipeline reproduces. C3 - PMD-026 reduces global accessibility (fewer peaks, lower total signal): direction reproduced. C1 - peak genomic distribution is dominated by intron+intergenic+TSS (canonical ATAC pattern reproduced); the exact ~75% headline is interpretation-dependent (intron+intergenic alone = 73.4% ~= 75%; the literal 3-category sum = 88.4%; difference attributable to Partek read-weighted pie vs our peak-count bedtools annotation) -> graded partial. NOT ATTEMPTED: Fig 5j/k/l GSEA + JASPAR motif (Partek Flow proprietary + R scripts on request) and all non-ATAC subseries. One real blocker hit and fixed: MACS2 __log_finite glibc symbol bug (LD_PRELOAD shim). No fabrication concern - every value independently recomputed from raw reads.
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.
Assessment versions
Every reproduction run is kept as an immutable version — anchored to the data as it stood, with a tamper-evident chain hash. A rerun (e.g. after an author updates a deposit) adds a new version; the previous one stays on record.
-
v1 current initial assessment Score 73assessed: 2026-06-20 ⛓ 7ad033f284cd
✎ 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
- 2026-06-20
- Rubric version
- v1.0
- Assessed by
-
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-20no human curator yet
- Last updated
- 2026-08-05
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: sonnetRSK1 (RPS6KA1) is a shared, exploitable oncogenic dependency across the spectrum of myeloproliferative neoplasms (MPN) and secondary AML (sAML) by driving NFκB-mediated hyperinflammation, and pharmacological targeting of RSK1 with the first-in-class inhibitor PMD-026 can suppress disease.
- ★ RSK1 is a conserved, exploitable therapeutic dependency across MPN and secondary AML finding
- ★ RSK inhibition with PMD-026 suppresses NFκB activation and pro-inflammatory mediators including TNF associated with disease severity/transformation finding
- ★ PMD-026 suppresses disease burden across seven syngeneic and patient-derived xenograft leukemia mouse models spanning diverse driver and disease-modifying mutations finding
- ★ A transcriptional (RNA-seq) and CyTOF atlas of 158 primary CD34+/CD14+ samples reveals aberrant PI3K/AKT/mTOR signaling and NFκB-mediated hyperinflammation across MPN/sAML finding
- ★ shRNA knockdown of RPS6KA1 reduces colony number/size in sAML CD34+ cells but not normal CD34+ cells finding
- ★ RPS6KA1 knockdown reduces leukemic engraftment (%hCD45) in a CD34+ sAML patient-derived xenograft mouse model finding
- ★ PMD-026 is a highly specific RSK1-4 inhibitor (99% specificity across 398 kinases) with low-nanomolar IC50s method
- YB-1 (YBX1) is a direct RSK1 substrate whose activity is inhibited by PMD-026 mechanism
| Assay | System | Perturbation | Readout | Platform |
|---|---|---|---|---|
| bulk RNA-seq | sorted CD34+ HSPCs from primary MPN/sAML patients and healthy NBM donors (WashU Cohort, n=90) | none | gene expression / DEGs / GSEA | — |
| targeted NGS myeloid gene panel / JAK2-CALR-MPL sequencing | primary MPN and sAML patient samples (n=73) | none | mutational landscape | — |
| mass cytometry (CyTOF) | CD34+ cells from MPN/sAML patients and normal donors (n=106) | none | phosphorylated and total protein expression (JAK-STAT, PI3K/AKT/mTOR/S6, NFκB, MAPK) | CyTOF |
| reverse phase protein array (RPPA) | 34 AML cell lines (CCLE) | none | correlation of pRSK1 with 54 phosphorylated effectors | RPPA |
| KINOMEscan kinase profiling | 398 kinases, cell-free/biochemical panel | drug (PMD-026) | kinase binding specificity, IC50 | KINOMEscan |
| cell viability assay | HEL cells; Ba/F3 cells ectopically expressing WT/mutant JAK2 and MPL | drug (PMD-026, BI-D1870) | dose-dependent viability/proliferation | — |
| Western blot / immunoblot | HEL cells | drug (PMD-026, 5-10 μM) | pRSK1, pS6, p-mTOR, YB-1 activity | — |
| CD34+ colony formation assay / shRNA knockdown xenograft (NSGS mice, intra-tibial) | primary MF/sAML patient CD34+ cells; PDX model (patient 784981) | drug (PMD-026) or shRNA RPS6KA1 knockdown | colony number/size; %hCD45 in blood/marrow, spleen/liver weight | MethoCult H4034 |
- – PCA of CD34+ RNA-seq shows a gradient of disease progression from NBM/PV/ET to MF to sAML
- ▲ mTOR/PI3K/AKT (RICTOR, AKT3, PIK3CA, PIK3C2A) and RAS/RAF (KRAS, BRAF) pathway gene expression is elevated in MPN/sAML vs NBM
- ▲ CyTOF shows hyperactivation of JAK-STAT, PI3K/AKT/mTOR/S6, NFκB, and MAPK signaling across MPN, most evident in sAML
- – PMD-026 shows specificity for RSK1-4 across 398 kinases tested by KINOMEscan 99% specificity
- – PMD-026 IC50 values determined for RSK1-4 2 nM (RSK1), 0.7 nM (RSK2), 0.9 nM (RSK3), 2 nM (RSK4)
- ▼ PMD-026 treatment suppresses pRSK1, pS6, p-mTOR, and YB-1 activity in HEL cells
- ▼ RPS6KA1 shRNA knockdown reduces colony number and size in sAML CD34+ cells but not normal CD34+ cells
- ▼ RPS6KA1 knockdown reduces %hCD45 engraftment in peripheral blood and bone marrow in a sAML PDX mouse model
- count 158 primary samples (CD34+ HSPCs and CD14+ monocytes) (WashU Cohort RNA-seq atlas)
- count 90 primary samples: PV n=6, ET n=9, MF n=30, sAML n=34, NBM n=11 (WashU Cohort CD34+ RNA-seq)
- count 106 primary samples: PV n=17, ET n=5, MF n=34, sAML n=15, normal donors n=35 (CyTOF cohort)
- fold_change >1 log2 fold change, 1810 candidate genes (sAML vs MF DEG-defined sAML transformation signature)
- other IC50: 2 nM (RSK1), 0.7 nM (RSK2), 0.9 nM (RSK3), 2 nM (RSK4) (PMD-026 potency counter-screen)
- other 99% specificity for RSK1-4 across 398 kinases (KINOMEscan profiling of PMD-026)
- count ASXL1 mutation prevalence: 23% (WashU sAML) vs 2.5% (TCGA) vs 10.3% (BeatAML2) (sAML vs de novo AML cohort mutation comparison)
- count TP53 23%, SF3B1 11.5% mutation prevalence in WashU sAML cohort (n=34) vs TCGA (n=200) and BeatAML2 (n=893) (sAML vs de novo AML cohort mutation comparison)
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 combines cross-sectional comparisons of gene expression (RNA-seq) and phosphoprotein signaling (CyTOF, RPPA) across disease subgroups (NBM/PV/ET/MF/sAML) with in vitro and in vivo functional experiments (viability, apoptosis, cell cycle, colony, and xenograft assays) evaluating RSK1 inhibition. Group comparisons are primarily assessed with two-tailed Student's t-tests and one- or two-way ANOVA with Dunnett's post-hoc correction, alongside Pearson correlation, RNA-seq differential expression (log2 fold-change based), and gene set enrichment analysis (GSEA). Results are reported mainly as means with error bars or boxplots, with figure-level sample sizes stated but no explicit power analysis, and significance is denoted with symbols in some panels rather than exact p-values within the text provided.
| Test | Applied to | n | Assumptions |
|---|---|---|---|
| two-tailed Student's t-test | Fig. 1e — expression of mTOR/PI3K/AKT and RTK/RAS/RAF pathway genes across disease groups (MF vs NBM, sAML vs NBM) | group sizes as in Fig. 1c: PV n=6, ET n=9, MF n=30, sAML n=34, NBM n=11 | not stated |
| two-tailed Student's t-test | Fig. 1g — CyTOF phosphoprotein/protein expression across disease groups (MF vs NBM, sAML vs NBM) | group sizes as in Fig. 1f: PV n=17, ET n=5, MF n=34, sAML n=15, normal donors n=35 | not stated |
| Pearson correlation | Fig. 2b — correlation of pRSK1 (T573, T359/S363) with 54 phosphorylated effectors from RPPA data | 34 AML cell lines (CCLE) | not stated |
| two-way ANOVA with Dunnett's multiple comparisons test | Fig. 2l — %hCD45 in peripheral blood across timepoints, shRPS6KA1 vs control xenograft mice | control n=4, sh#1 n=5, sh#2 n=5 | not stated |
| one-way ANOVA with Dunnett's multiple comparisons test | Fig. 2l — %hCD45 in bone marrow and normalized spleen/liver weights, shRPS6KA1 vs control xenograft mice | control n=4, sh#1 n=5, sh#2 n=5 | not stated |
| gene set enrichment analysis (GSEA) / DEG analysis (log2 fold-change threshold) | Fig. 2i and sAML transformation signature analysis — RNA-seq of PMD-026-treated HEL cells and CD34+ sAML vs MF HSPCs | not stated | na |
-
Gene and phosphoprotein expression across multiple disease groups (NBM, PV, ET, MF, sAML) were compared using pairwise two-tailed Student's t-tests (Fig. 1e, 1g).↳ Could also: A one-way ANOVA across all disease groups with a post-hoc test (e.g., Tukey's or Dunnett's) could also be used. — When more than two groups are being compared on the same outcome, an omnibus ANOVA with post-hoc correction jointly evaluates all group differences and helps control the family-wise error rate that can arise from performing several separate pairwise t-tests.
-
Variability throughout the figures is summarized as mean values ± SD, including for assays with small numbers of biological replicates (e.g., n=3).↳ Could also: Reporting the SEM or a 95% confidence interval alongside or instead of SD would also convey the precision of the estimated mean. — For small-n experiments, SD mainly reflects sample spread, whereas SEM/CI communicates the uncertainty of the mean estimate itself, which some readers find more informative when judging reproducibility of an effect.
-
Multiple genes/proteins were each tested individually with t-tests within the same figure panel (e.g., Fig. 1e comparing several pathway genes across disease groups).↳ Could also: A multiple-testing correction such as Benjamini-Hochberg FDR or Bonferroni adjustment across the full panel of genes/proteins tested could also be applied. — Testing many genes within one figure increases the chance of some comparisons reaching significance by chance; an FDR or Bonferroni adjustment is a standard way to account for this when many hypotheses are evaluated together.
-
Association between pRSK1 markers and 54 phosphorylated effectors in RPPA data was assessed using Pearson correlation (Fig. 2b).↳ Could also: A Spearman rank correlation could also be used for this analysis. — Spearman correlation does not assume a linear relationship or normally distributed data and is less sensitive to outliers, which can be useful for phosphoprotein signal data that may not be normally distributed.
-
The sAML transformation signature was defined using a log2 fold-change threshold (>1) between sAML and MF HSPCs.↳ Could also: Combining the fold-change threshold with an adjusted p-value/FDR cutoff (as in standard DESeq2/edgeR/limma workflows) could also be used to define the gene signature. — Using effect size (fold-change) together with a significance/FDR threshold jointly accounts for both the magnitude and statistical reliability of expression differences, which is a common convention in RNA-seq differential expression analysis.
-
Comparisons involving small biological replicate numbers (e.g., n=3 for colony assays, apoptosis, and cell cycle assays) were assessed with parametric tests (t-test/ANOVA).↳ Could also: Non-parametric alternatives such as the Mann-Whitney U test or Kruskal-Wallis test could also be used for these comparisons. — Non-parametric tests do not rely on an assumption of normally distributed data, which can be difficult to verify with very small sample sizes, and are sometimes preferred in that setting.
What was reproduced
The exact results taken into scope, with each reported value next to the value our attempt produced.
Scope — pmid-39820365
Paper: Kong T, Laranjeira ABA, Letson CT, et al. "RSK1 is an exploitable dependency in myeloproliferative neoplasms and secondary acute myeloid leukemia." Nat Commun 2025;16:xxx. PMID 39820365 · PMCID PMC11739599 · DOI 10.1038/s41467-024-55643-7.
What kind of paper
Predominantly a wet-lab cancer-biology paper (drug screens with PMD-026/RSK inhibitor, mouse models, CyTOF, flow, biochemistry). A minority of the figures are sequencing-pipeline-derived. Per HARD RULE 2 we attempt only the pipeline-derived results and explicitly do NOT attempt the wet-lab work.
Sequencing datasets generated by THIS paper (verified via NCBI elink PMID→GEO)
SuperSeries GSE228760 (BioProject PRJNA953424), sub-series:
| Sub-series | Assay | n | Pipeline named in Methods |
|---|---|---|---|
| GSE283710 | bulk RNA-seq, CD34+ HSPC | 90 | (RNA-seq aligner; analysis in R, scripts "on request") |
| GSE283711 | bulk RNA-seq, CD14+ mono | 68 | same |
| GSE228758 | bulk RNA-seq, cell lines (HEL/THP-1 + inhibitor) | 16 | same |
| GSE228759 | scRNA-seq (patient 551599) | 2 | (10x / Seurat) |
| GSE229220 | ATAC-seq, THP-1 ±PMD-026 | 6 | ENCODE-DCC/atac-seq-pipeline v2.2.0 → hg38 |
| GSE229174 | CUT&Tag H3K27ac, THP-1 | 9 | ENCODE-DCC/chip-seq-pipeline2 v2.1.6 → hg38 |
NOTE on the scaffold accession GSE135902: it is a text-mining false positive — it is "The Transcriptome of CMML Monocytes" (Hsu et al., PMID 31648319, 2019), used by this paper only as an external reference cohort for a Fig. 6e comparison. It is NOT this paper's own data. The paper's own ATAC data is GSE229220. Likewise the scaffold "code" link
ENCODE-DCC/atac-seq-pipelineis a third-party tool, not the authors' own repo — exactly the P16 case.
Code availability (verbatim from the paper)
"No new code was developed in this study. R scripts utilized in this study are available from the corresponding author upon request."
Consequence: the authors' downstream analysis code is not public. The only publicly runnable, named pipeline is the upstream ENCODE ATAC-seq pipeline v2.2.0 applied to GSE229220 — which is exactly what we reproduce (third-party tool on the paper's own data = a fully valid reproduction, HARD RULE 2 / P16).
IN SCOPE (attempted) — ATAC-seq, GSE229220
Methods (verbatim): "samples were demultiplexed, underwent adapter trimming, filtering, and alignment to the hg38 reference genome following the ENCODE ATAC-seq pipeline (github.com/ENCODE-DCC/atac-seq-pipeline, v2.2.0). … peaks were generated with MACS, and regions were annotated to hg38." 6 samples = THP-1 DMSO ×3 (SRR24107899/898/897) vs PMD-026 5 µM 24 h ×3 (SRR24107896/895/894).
Reproduction strategy: run the ENCODE ATAC-seq pipeline's exact algorithm and parameters (bowtie2 → hg38, MAPQ/dup/chrM/blacklist filtering, MACS2 ATAC peak calling with Tn5 shift, MACS2 fold-enrichment signal) on the paper's own raw FASTQ, on «our HPC»/«infra». Compare to:
- Claim C1 (Fig. 5h,i): "intronic, intergenic, and TSS regions encompassing ~75% of reads." → reproduce by annotating MACS2 peaks to hg38 gene features and summing promoter(TSS)+intron+intergenic fractions.
- Claim C2 (deposited processed output): the authors deposited two pooled
fold-change signal bigWigs (
GSE229220_DMSO_atac.fc_signal.bigwig,..._PMD_atac.fc_signal.bigwig). → reproduce pooled-per-condition MACS2 FE signal and compute genome-wide correlation against the deposited tracks. - Claim C3 (qualitative, Results/Discussion): "a reduction in transcriptional accessibility by PMD-026." → check global accessibility (peak count / total signal) DMSO vs PMD direction.
OUT OF SCOPE (not attempted) — and why
- Fig. 5j,k,l (GSEA of differential peaks; differential TSS regions; JASPAR motif enrichment): performed in Partek Flow (proprietary GUI) + author R scripts available "on request" → not publ
Assessments & scoring basis
Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.
An automated assessment. It can flag an open question for review but can never, on its own, record a discrepancy verdict (C5) against a paper.
Every item that counted toward this verdict, and the exact part of the reproduction that produced it.
Strong, fair reproduction on the paper's own raw ATAC data (GSE229220): the quantitative anchor C2 matches the authors' deposited fold-change track at r≈0.90 genome-wide and C3's direction (PMD-026 lowers accessibility) is confirmed, with no fabrication concern since every value was recomputed from FASTQ. The deviations are on our/method-definition side: C1's ~75% headline is interpretation-dependent (literal sum 88.4% vs intron+intergenic-only 73.4%≈75%, a Partek read-weighted-pie vs peak-count difference), and Fig.5j,k,l GSEA/motif is unreproducible due to proprietary Partek Flow. Overall yellow: solid reproduction with explainable, non-substantive deviations and one availability-gated skip — the central ATAC conclusions hold.
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
<synthetic>Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.