Corpus 1,272 assessed · 1,173 scored · 643 reproduced ≥75 · 168 flagged ·∅ 74.1/100
← New search

A feed-forward pathway drives LRRK2 kinase membrane recruitment and activation.

Elife · 2022
L1 50/100 3/4
Why this verdict

The main results reproduced, with only marginal, non-material deviations.

Reproduced on the brainbox compute brainarbeit.com
✓ What held up
  • Reported values were directly comparable
What did not (or only partly)
  • 🟡Could not use the authors’ exact input data
  • 🟡A deviation arose in the data or preprocessing
  • 🟡A deviation was attributed to the published material
  • 🟡Reported values were not (fully) derivable from the shared data
  • 🟡The deviation was non-trivial in magnitude
  • 🟡The central claim did not (fully) hold under reproduction
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
50/100
Reproducibility score
1.4 SD below mean
vs. all fields · 1173 studies
🎯 Scores higher than 8% of all assessed papers rank 1026 of 1173 scored

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

IN PROGRESS. Target = R/TIRF single-molecule tracking pipeline (Rmd/rab10_tirf.Rmd -> Fig 8 fig-supp 1). Reported track counts at t=0: wt 1171, wt_rep2 2334, d2017a2 50, wt_PDL 106. paperutils deps are cosmetic (stubbable). OPEN: input track CSVs (data-raw/*.csv) are NOT in the repo nor the 602KB Zenodo code zip; must verify whether they live inside Dryad Figure_8A_B_C.zip (3.26GB). «our HPC» SSH currently timing out (not touching VPN; waiting for central fix). Not yet attempted: AlphaFold/ColabFold structural predictions (Fig 3A/5B).

💻 Code ↗ 🗄 Data: 10.5281/zenodo.7057419

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.

  1. v1 current initial assessment Score 50
    assessed: 2026-06-19 ⛓ 498c7ee6faba
✎ 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.

Reason for the rerun

We email you a confirmation link first. The rerun is an objective re-measurement — it cannot change the verdict in your favour, only ask us to look again.

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
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-19
no human curator yet
Last updated
2026-07-31

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: opus
Founding hypothesis

How does membrane recruitment activate LRRK2 kinase, and do LRRK2-phosphorylated Rab GTPases create a feed-forward mechanism that further recruits and activates LRRK2 on membranes?

Core claims
  • A C-terminal patch of the LRRK2 Armadillo domain (residues ~350–550, 'site #1') binds non-phosphorylated Rab29, Rab8A, and Rab10 with low-micromolar affinity finding
  • A distinct N-terminal interface of the LRRK2 Armadillo domain (residues 1–159, 'site #2') binds specifically and with higher affinity to LRRK2-phosphorylated Rab8A and Rab10 finding
  • LRRK2 residues Arg361, Arg399, Leu403, and Lys439 form the site #1 Rab-binding surface; K439E is recommended to block Rab29 interaction and activation mechanism
  • N-terminal basic residues Lys17 and Lys18 are required for phosphoRab binding at site #2 mechanism
  • PhosphoRab binding establishes a feed-forward pathway that retains and further activates LRRK2 on membranes, providing spatial control of kinase activity mechanism
  • Rapid recovery of LRRK2 kinase activity after inhibitor washout depends on LRRK2 association with phosphorylated Rab proteins, and phosphoRab8A stimulates LRRK2 phosphorylation of Rab10 in vitro finding
  • Only active LRRK2 cooperatively associates with phospho-Rab10-decorated planar lipid bilayers finding
  • An MST3 kinase-based in vitro Rab phosphorylation method enables monitoring phosphoRab binding via microscale thermophoresis method
Experimental setups
Assay System Perturbation Readout Platform
Microscale thermophoresis (MST) binding Purified recombinant LRRK2 Armadillo domain fragments and Rab GTPases (Rab29, Rab8A Q67L, Rab10 Q68L, Rab7) Armadillo fragment truncations (1–552, 1–159, 350–550) and point mutants (K17A, K18A) Binding affinity (KD) NHS-RED labeling; Mg2+-GTP conditions
Microscale thermophoresis with phosphoRabs Purified LRRK2 Armadillo fragments with MST3-phosphorylated Rab8A Q67L and Rab10 Q68L In vitro phosphorylation by MST3 kinase; Armadillo K17A/K18A mutants PhosphoRab binding affinity (KD) MST3 kinase phosphorylation (27°C, 2 hr); NHS-RED labeling
Confocal immunofluorescence microscopy / colocalization HeLa cells co-expressing GFP-LRRK2 (full-length and fragments/mutants) and HA-Rab29 LRRK2 fragment truncations and point mutations (R361E, R399E, L403A, K439E) LRRK2–Rab29 Golgi colocalization (Mander's coefficient) CellProfiler software
Immunoblotting of Rab phosphorylation (cellular kinase activity) HEK293T cells co-expressing GFP-LRRK2 (WT or R1441G) and HA-Rab29 Rab29 overexpression; LRRK2 site #1 point mutants; pathogenic R1441G pRab10 levels (phospho-Rab10 normalized to Rab10) LI-COR (800/680 channels), ECL; anti-pRab10, anti-Rab10, anti-HA
In silico structural modeling/docking LRRK2 350–550 fragment with Rab29 and Rab8A; full-length LRRK2 Armadillo domain none Predicted interface residues, electrostatic surface potential AlphaFold/ColabFold (AlphaFold2_advanced.ipynb), ChimeraX 1.4, APBS, ConSurf
Kinase inhibitor washout (activity recovery) Cells expressing LRRK2 Kinase inhibitor addition then washout Recovery of LRRK2 kinase activity dependent on phosphoRab association
In vitro kinase assay Purified LRRK2 with phosphoRab8A and Rab10 Pre-phosphorylated Rab8A LRRK2-mediated Rab10 phosphorylation rate
Reconstitution on planar supported lipid bilayers Purified LRRK2 on bilayers decorated with Rab10 protein Active vs inactive LRRK2; phospho-Rab10 vs Rab10 Cooperative LRRK2 membrane recruitment/association
Key results
  • Rab29 binds full-length Armadillo (1–552) and 350–550 fragment but not 1–159 KD = 1.6 µM (1–552 and 350–550); >29 µM for 1–159
  • Non-phosphorylated Rab8A and Rab10 bind site #1 (350–550) and full Armadillo Rab8A KD 2.9 µM (1–552), 2.3 µM (350–550); Rab10 KD 2.4 µM (1–552), 5.1 µM (350–550)
  • PhosphoRab8A and phosphoRab10 bind tightly to N-terminal 1–159 (site #2) but not to 350–550 pRab8A KD ~1.0 µM, pRab10 KD ~0.71 µM at 1–159; >19–29 µM at 350–550
  • K439E site #1 mutation completely blocks Rab29-mediated activation of pathogenic R1441G LRRK2; R399E shows weak activation
  • Single mutation of K17 or K18 abolishes phosphoRab10 binding to the Armadillo domain KD increased to >20 µM
  • K18 is highly conserved (ConSurf score 8) versus K17 (score 2) ConSurf score 8 vs 2 (max 9)
  • Non-LRRK2 substrate Rab7 fails to bind the Armadillo 1–552 fragment KD >6.5 µM
  • Mutations R361E, R399E, L403A, K439E block LRRK2–Rab29 Golgi colocalization in HeLa cells
Key statistics
  • other KD = 1.6 ± 0.9 µM (Rab29, Armadillo 1–552) (Rab29 binding to full Armadillo by MST)
  • other KD = 1.6 ± 0.5 µM (Rab29, Armadillo 350–550) (Rab29 binding to site #1 fragment)
  • other KD = 0.71 ± 0.3 µM (pRab10, Armadillo 1–159) (phosphoRab10 binding to site #2)
  • other KD = 1.0 ± 0.6 µM (pRab8A, Armadillo 1–159) (phosphoRab8A binding to site #2)
  • other KD = 2.9 ± 1.2 µM (Rab8A, Armadillo 1–552) (non-phospho Rab8A binding)
  • other KD = 2.4 ± 0.6 µM (Rab10, Armadillo 1–552) (non-phospho Rab10 binding)
  • other pRab10 KD >20 µM upon K17A or K18A mutation (loss of phosphoRab10 binding at site #2)
  • other McGrath et al. reported Rab29/32/38 affinities 2.7, 1.2, 1.2–2.4 µM (prior reported affinities for comparison)

Statistical methods review

Model: sonnet

A 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 reports in vitro biophysical binding assays (microscale thermophoresis) between LRRK2 Armadillo domain fragments and various Rab GTPases (native and phosphorylated), summarizing dissociation constants (K_D) as mean ± SEM from three independent measurements per condition. Cell-based experiments quantify LRRK2-Rab29 co-localization using a Mander's coefficient (CellProfiler) and immunoblot band intensities reported as mean ± SD of duplicate determinations. No inferential hypothesis tests, p-values, or formal multiplicity corrections are described in the text provided.

Replicationmixed Sample sizeStated per experiment as number of independent replicates (e.g., n=3 independent measurements from different protein preparations for MST assays; duplicate, n=2, determinations for immunoblot quantification); no formal sample-size/power calculation is described. GroupsBinding affinities and activation/co-localization signals compared across LRRK2 constructs (wild-type vs. site-directed point mutants, e.g., R361E, R399E, L403A, K439E, K17A, K18A) and across Rab GTPases (phosphorylated vs. non-phosphorylated Rab8A/Rab10/Rab29/Rab7, and Armadillo sub-fragments). Pairingna Randomization/blindingnot stated Dispersionmixed Exact p-valuesno Effect sizesno Confidence intervalsno
Statistical tests used
Test Applied to n Assumptions
Descriptive summary statistics (mean ± SEM) from curve-fitting of binding curves Microscale thermophoresis K_D determinations (Figures 1, 2, 4, 5; Table 1) three independent measurements, each from a different protein preparation not stated
Descriptive summary statistics (average ± SD) of quantified immunoblot signals GFP-LRRK2/Rab29 co-expression activation assays (Figure 3C and D) duplicate determinations not stated
Mander's overlap coefficient (image co-localization metric) GFP-LRRK2/HA-Rab29 co-localization at the Golgi (Figure 3B) not stated not stated
Approaches that could also have been used
  • K_D values for wild-type versus mutant LRRK2/Rab interactions are compared as point estimates (mean ± SEM, n=3) without a stated formal statistical test.
    Could also: An extra sum-of-squares F-test or similar model-comparison test built into nonlinear regression/curve-fitting software could also be applied to formally test whether K_D values differ between conditions. — This would provide a formal statistical basis (e.g., a p-value) for concluding that two binding curves/affinities differ, complementing the visual comparison of point estimates and error bars.
  • Binding and quantification data from small replicate numbers (n=3 for MST assays, duplicates for immunoblot quantification) are summarized using SEM or SD.
    Could also: Reporting a 95% confidence interval alongside or instead of SEM/SD would also convey the precision of these small-sample estimates. — With very small n, SEM can visually understate variability, and a CI directly communicates the range of plausible true values for the estimate.
  • Effects of many individual LRRK2 point mutants on Rab29 co-localization and kinase activation are compared largely descriptively across a panel of constructs relative to wild-type.
    Could also: A one-way ANOVA with a post-hoc test such as Dunnett's (comparing each mutant to the wild-type control) could also be used across the mutant panel. — This approach formally quantifies which mutants differ from wild-type while controlling the family-wise error rate across the multiple comparisons being made.
  • Immunoblot-based activation measurements in Figure 3C and D are based on duplicate (n=2) determinations.
    Could also: Increasing to three or more independent biological replicates and applying a standard t-test or ANOVA could also be used for these comparisons. — A larger n improves the reliability of variability estimates (SD/SEM) and supports more robust inferential statistical comparisons where desired.
  • Visual scoring of LRRK2 fragment co-localization with Rab29 in Figure 3—figure supplement 1 is described as assessed 'visually' by microscopy.
    Could also: Blinded or automated/computational image classification could also be used to score co-localization. — Blinding or automation of the scoring step can reduce the potential for observer expectation to influence categorical co-localization calls.
  • Binding curve fitting (K_D determination) is performed without naming the specific curve-fitting/statistical software or version used.
    Could also: Explicitly citing the curve-fitting software and version (e.g., a specific MST analysis package or GraphPad Prism) could also be reported. — Naming the specific software and version supports reproducibility and lets readers evaluate the fitting model and error-estimation approach used to derive K_D values.
Software: CellProfiler · ColabFold / AlphaFold2_advanced.ipynb · ChimeraX 1.4 · Consurf server

What was reproduced

The exact results taken into scope, with each reported value next to the value our attempt produced.

Scope — pmid-36149401

Paper: Vides EG et al. (2022) A feed-forward pathway drives LRRK2 kinase membrane recruitment and activation. eLife 11:e79771. PMID 36149401.

Code: https://github.com/PfefferLab/Vides_et_al_2022 (MIT, MATLAB+R, last push 2022-09-23, commit 2b50525ee1d48790466d35222956f16615ae96e8; Zenodo archive 10.5281/zenodo.7108943 = same code zip).

Primary data: Dryad 10.5061/dryad.3tx95x6j7 (v5, ~5.86 GB, 9 figure zips + README). Zenodo 10.5281/zenodo.7057419 = Fig3-S4 immunoblot images + xlsx.

Classification of reported results

Result Origin Pipeline In scope?
Fig 8 fig-supp 1 — single-molecule TIRF tracking of Rab10: intensity evolution, track persistence/decay, initial-intensity distribution, % ultrabright (multimer) computational Rmd/rab10_tirf.Rmd (R 4.2: tidyverse, ggridges, cowplot, lemon, here, paperutils) reading TrackIt-exported track CSVs YES — primary target
Track export from TrackIt .mat → CSV computational tracks_export.m (MATLAB) partial (needs raw .mat; CSVs are the pipeline input)
Fig 3A, Fig 5B — AlphaFold2/ColabFold LRRK2–Rab complex models + contact residues computational ColabFold/AlphaFold2 (GPU) secondary — heavy, qualitative comparison only
Fig 6 — Mander's colocalization coefficients computational CellProfiler (GUI pipeline, params not fully specified) out (no shipped CP pipeline file; GUI)
Fig 8A/B/C recruitment curves wet-lab + Prism GraphPad Prism nonlinear regression (manual) out (not a bioinformatic pipeline)
Figs 1,2,4,7,9 — MST, immunoblot, kinase assays wet-lab out

Primary reproduction target: rab10_tirf.Rmd

A self-contained, version-controlled R analysis. The repo ships the rendered output Rmd/rab10_tirf.md with 8 committed figure PNGs and printed numbers, so the reproduction is a true self-consistency check (re-run → compare to committed output) and a comparison to the paper's Fig 8 fig-supp 1.

Reported numeric outputs to match (from committed rab10_tirf.md): tracks present at t=0 — wt 1171, wt_rep2 2334, d2017a2 50, wt_PDL 106. Plus: intensity-evolution ridge plots (log2 I/I0), exponential persistence decay, % ultrabright (log2(Ainit/Ainit0) > 1.5) over time.

Inputs: data-raw/{rab10rgnew_tracks.csv, pdlrg_tracks.csv, d2017a2_tracks.csv, WT_tracks.csv} — NOT committed to the repo, NOT in the Zenodo code zip (602 KB, too small). Must be located inside Dryad Figure_8A_B_C.zip (3.26 GB, Figure8RAWdata) — to be confirmed by listing the zip on front1. If absent there, the analysis input is undeposited → that result becomes data_unavailable (honest partial).

Env note: paperutils is a custom (non-CRAN, climouse/Limouse) package. Only two functions are used and both are cosmetictheme_publish() (ggplot theme) and prettysave() (ggsave wrapper) — so they are stubbed without affecting any numeric result. Env is resolvable.

Compute

All heavy compute on «our HPC» («infra») via «host» ssh «host». Downloads on front1 → «infra» work dir «path». R env via conda on front1.

Figures / tables: Fig 8
c1_ntracks_wt
Reported
1171
Reproduced
partial
c2_ntracks_wt_rep2
Reported
2334
Reproduced
partial
c3_ntracks_d2017a2
Reported
50
Reproduced
partial
c4_ntracks_wt_pdl
Reported
106
Reproduced
partial

Assessments & scoring basis

Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.

🤖 AI curator · claude (ai-curator room) · v1.0 L1 50/100

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.

🟡1. Data identity
🟢2. Endpoint comparability
🟡3. Location of the main deviation
🟡4. Cause of the deviation
🟡5. Derivability / plausibility
🟡6. Severity of the deviation
🟡7. Core claim
🟡8. Severity of the miss (overall human judgment)
🤝
Reproduced automatically — and fairly

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-4-8

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.

92.3 k
tokens (I/O) · 4.1 M incl. cache
12 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.