Can public evidence prioritize testable drug-repurposing hypotheses for BUB1B L737* and N1002K?
The reading order of the MVA · BUB1B case as a trail: the two variant questions, the evidence records behind each gate, the three stated hypotheses, the gaps that keep them conditional, and the measurements proposed next. Every step links to the records, publications and graph nodes it rests on; nothing is concluded beyond what they say.
Curated: Case page reading order: overview → 01 Summary and findings → 02 Questions → evidence records → 05 October 24 revision → 06 Future validation · updated
01
Question
Two BUB1B variants, two molecular questions
L737* introduces a premature stop codon, an RNA and translation question. N1002K is a missense change in the kinase region, a protein stability and function question. Both sit in BUB1B, which encodes the spindle assembly checkpoint component BUBR1, in a case of mosaic variegated aneuploidy. The guiding question of the case: can public evidence prioritize testable drug-repurposing hypotheses?
Source: Case page, overview ("The question", "Case at a glance")
02
Evidence
The RNA gate for L737*
Three records organise the RNA question. The NMD-first audit finds a related 731X construct that motivates testing whether a truncated product retains function, while the exact L737* experiment has not been performed. The readthrough reanalysis finds no exact L737* context in the 5,837-PTC table. The transfer audit shows that endogenous-protein readthrough evidence comes from another gene and stop context.
Source: Case page, question "RNA availability"; records MVA-L737-NMD-FIRST-001, MVA-RT-002, MVA-RT-TRANSFER-009 and the publications they cite
03
Hypothesis
Preserving RNA leaves the stop codon
Stated hypothesis of the case: NMD modulation alone cannot establish a useful protein product; readthrough needs exact-context and functional validation. It remains proposed. The gaps it depends on are listed with it.
Source: Case page, 01 Summary and findings (card "L737* / RNA & translation")
04
Evidence
The function gate for N1002K
The equal-expression audit locates no exact-N1002K functional experiment; comparator variants make an abundance-limited mechanism plausible, but equally expressed mutants elsewhere in the protein still failed chromosome congression. The arimoclomol audit finds no target engagement or exact-variant rescue. The arrest-duration reanalysis is a checkpoint control (WT 650 min, D137R 60 min), not a drug result.
Source: Case page, question "Protein function"; records MVA-N1002K-GATE-002, MVA-ARI-007, MVA-CTRL-006 and the publications they cite
05
Hypothesis
Abundance is a starting point
Stated hypothesis of the case: stability and residual function remain unresolved; matched protein abundance must be followed by independent functional measurements. It remains proposed.
Source: Case page, 01 Summary and findings (card "N1002K / protein function")
06
Evidence
Candidates, under conditions
The 704-drug NMD screen reanalysis names montelukast as a reporter-dependent hit with no functional BUBR1 rescue. The amlexanox audit finds functional evidence in one gene and negative evidence with toxicity in another. The rapamycin reanalysis finds the BUB1B locus RNA signal 41% lower, which does not support an RNA-increase rationale. The immunotherapy audit separates tumor control from disease rescue.
Source: Records MVA-NMD-003, MVA-AMLEXANOX-010, MVA-RAP-GSE65145-001, MVA-IMMUNE-011 and the publications they cite
07
Hypothesis
Research candidates stay conditional
Stated hypothesis of the case: amlexanox and montelukast remain conditional, gentamicin plus mefloquine remains an experimental branch, and no exact-genotype pharmacological rescue is established.
Source: Case page, 01 Summary and findings (card "Research candidates")
08
Evidence
Robustness and limits (October 24, 2026)
A post-hoc exploratory computational audit, published in the research repository, re-read two earlier observations. Readthrough: the nearest set is 12 measured variant rows representing 8 unique six-base contexts, with no exact L737* context; the median paired increment stays positive under every row and context-group deletion. Rapamycin: the BUB1B-spanning RNA feature stays below the proliferative control under every specified feature, estimator and sample check. These are finite diagnostics, not confidence intervals, biological experiments or efficacy; proliferation confounding is unresolved. The candidate decision matrix is unchanged: no candidate is promoted.
Source: Case page, 05 October 24 revision: "Robustness and limits" and "Candidate decision matrix"
09
Evidence gap
What remains unresolved
Exact-variant function. L737* NMD susceptibility is predicted, not measured. N1002K stability and residual function are unresolved. Checkpoint competence and chromosome segregation have not been measured for either variant.
Source: Case page, overview: "What remains unresolved"
10
Proposed next step
The next decisive measurements
Proposed work, not performed by this project: (01) establish L737* product function and N1002K activity at matched protein abundance; (02) measure checkpoint competence and chromosome segregation separately; (03) account for viability, proliferation and completed divisions before interpreting a drug response.
Source: Case page, 06 Future validation: "The next decisive measurements"
Evidence notes 0
Researchers' annotations on this item. They are the researcher's reading, not the content of a publication.
No notes yet.
BioLab is the biomedical research area of Stackfire Labs: research cases built around open evidence, reproducible reanalyses and an explicit record of what remains unresolved. It organizes evidence and proposes experiments; it does not run patient experiments or recommend treatments.
Two germline BUB1B variants in mosaic variegated aneuploidy raise separate questions: L737* is an RNA and translation question, N1002K a protein stability and function question. The case records reanalyses and evidence audits, an October 24, 2026 robustness audit and a candidate decision matrix. No exact-genotype pharmacological rescue is established; wet-lab validation remains future work.
A shared research library of publications kept as references, not copies. Every entry links to its original source and records which public bibliographic services supplied its metadata and when. Reading states and collections are bibliographic organization, never scientific validation.
The graph connects a case to its variants, processes, evidence records, publications and hypotheses. Solid lines are documented connections; dashed lines are hypotheses under review. A connection says that a source or record links two items: it is not evidence of causation or efficacy. Every view also has a list alternative.
A research trail is an ordered reading of the evidence, step by step. Hypotheses are statements under research with the evidence that informs or limits them; none is marked validated. Evidence gaps name what has not been measured or located. All of them link to existing records and publications instead of copying them.
The auditable record of the MVA reanalyses with their execution records and downloads, and a Protein Structure Explorer of the predicted BUBR1 model. Tools marked Planned are not built yet. Reanalyses of public processed data and predicted structures are not biological experiments.
No. BioLab is research context, not a clinical diagnostic report or a treatment recommendation. Drug candidates are research hypotheses with explicit gates, not proven treatments. Questions about a person’s care belong with their clinical team.