Analyses of the case
Each analysis page shows what was archived, what was executed and with which record, where every number comes from, and what the result is not. Case page: MVA · BUB1B.
- Archived result
- Executed reanalysis
- Exploration
The four reanalyses
Original deterministic reanalyses from processed public sources, archived in the October 17 snapshot and replayed with execution records.
-
Rapamycin in human fibroblasts: BUB1B locus RNA signal
- Archived result
- Executed reanalysis
What it is not. An RNA feature in normal fibroblasts, not BUBR1 protein, not an MVA model and not a rescue. Reduced serum is not a cell-cycle-matched control; proliferation confounding is unresolved.
Explore results -
704-drug NMD screen: reporter specificity
- Archived result
- Executed reanalysis
What it is not. A reporter screen in mouse cells. No BUB1B was measured; statistical significance is not therapeutic suitability and no hit establishes functional BUBR1 rescue.
Explore results -
L737* readthrough: exact versus nearby contexts
- Archived result
- Executed reanalysis
What it is not. There is no exact L737* measurement and no exact six-base match. Twelve measured rows represent eight distinct contexts; they are not BUB1B efficacy, not a patient prediction and not a rescue probability.
Explore results -
BUBR1 functional controls: arrest duration
- Archived result
- Executed reanalysis
What it is not. A control for checkpoint competence, not a drug-rescue result. No drug and no N1002K were tested.
Explore results
Sensitivity audit
The October 24 post-hoc exploratory audit of two of them.
-
Post-hoc sensitivity audit (October 24, 2026)
- Archived result
- Executed reanalysis
What it is not. Post-hoc and exploratory. Diagnostic ranges are not confidence intervals; nothing here increases n or establishes biological validation.
Explore results
Structure metrics
Geometry of the predicted BUBR1 model around residues 1002 and 1012, reproduced in September.
-
BUBR1 AlphaFold model: residues 1002, 1012 and the L737* transcript context
- Archived result
- Executed reanalysis
- Exploration
What it is not. Predicted-model geometry. pLDDT is model confidence, not evidence of function; nothing here is a pathogenicity call or a drug-binding result.
Explore results