MVA Rare Disease Lab
Reproducible genomic analysis for a real child with mosaic variegated aneuploidy. Our Track 1 proposal at the MVA Hackathon 2026 by Sage Bionetworks prioritised a compound heterozygous pair in BUB1B, and it earned the highest score in the track.
Research hypothesis submitted to a hackathon, not a clinical diagnostic report. Variant interpretation and phase confirmation require appropriate clinical validation before any medical use.
compound heterozygous · autosomal recessive model
BubR1 · 1050 aa
VIRTUAL LAB / RESULTS
Evidence you can inspect.
Recorded analyses, their limits, and the next experiment that could change the decision.
Research, not a treatment recommendation. Public literature, computational reanalyses and explicitly labeled synthetic software checks only. Wet-lab validation is deferred. No private genomic data are included. The GSE65145 summary and N1002K equal-expression audit were checked by the research assistant for source support, privacy and scientific limits; neither has received independent human scientific review. “Completed” describes the recorded computational or review work, not biological validation. No exact-genotype pharmacologic rescue is established in this snapshot.
4Reanalysisrecords in this snapshot
3Evidence auditrecords in this snapshot
1Software testrecords in this snapshot
0Proposed experimentrecords in this snapshot
8 records. Expand a record to inspect evidence and provenance.
| Record / question | Work & execution | Finding | Decision | Recorded on |
|---|---|---|---|---|
| MVA-NMD-003704-drug NMD screen: reporter specificity | ReanalysisCompleted · COMPUTATIONAL / PATHWAY EVIDENCE | Montelukast has a published permutation q of 0.021150, but the response is reporter-dependent and confounded by non-NMD isoform changes. No hit establishes functional BUBR1 rescue. | Conditional | |
| MVA-RT-002L737* readthrough: exact versus nearby contexts | ReanalysisCompleted · COMPUTATIONAL / PATHWAY EVIDENCE | No exact L737* measurement or exact six-flanking-base match in the 5,837-PTC table. Twelve nearest contexts have a median paired gentamicin increment of 0.0165 percentage points. | Deprioritized | |
| MVA-CTRL-006BUBR1 functional controls: arrest duration | ReanalysisCompleted · COMPUTATIONAL / DIRECT-BUBR1 | Figure 4B source-data medians are 650 minutes for WT and 60 minutes for D137R. A control for checkpoint competence, not a drug-rescue result. | Control only | |
| MVA-N1002K-GATE-002N1002K stabilization: equal-expression gate | Evidence auditCompleted · DIRECT-BUBR1 COMPARATORS / RELATED-VARIANT IN VIVO / TARGETED AUDIT | No exact-N1002K functional experiment was located. I909T/L1012P checkpoint rescue at matched abundance makes an abundance-limited mechanism plausible, but equally expressed B56-binding mutants still failed chromosome-congression rescue and mouse L1012P showed allelic effects beyond total protein level. Stabilization remains conditional; no drug is promoted. | Conditional | |
| MVA-L737-NMD-FIRST-001L737* NMD-first: test truncated-protein function | Evidence auditCompleted · DIRECT-BUBR1 COMPARATOR / OUR INFERENCE | A related 731X construct motivates testing whether L737* retains useful function when NMD is bypassed. The exact L737* experiment has not been performed. | Conditional | |
| MVA-ARI-007Arimoclomol: target engagement and exposure gap | Evidence auditCompleted · PATHWAY EVIDENCE / COMPUTATIONAL / OUR INFERENCE | No direct BUBR1 or N1002K rescue evidence was located in the audit. Heat-shock responses are context-dependent; arimoclomol is not promoted to lead candidate. | Deprioritized | |
| MVA-FUNC-005Cytostasis confounding: denominator unit test | Software testCompleted · SYNTHETIC SOFTWARE FIXTURE | The synthetic fixture flags fewer raw error-positive divisions without a lower error rate per completed division. It does not measure sirolimus activity. | Control only | |
| MVA-RAP-GSE65145-001Rapamycin in human fibroblasts: BUB1B locus RNA signal | ReanalysisCompleted · Public human perturbation data; descriptive locus-level RNA reanalysis, not functional rescue. | The primary BUB1B-spanning feature has a Rap/Pro mean ratio of 0.585706 (41.43% lower); reduced-serum/Pro is 0.186191. Three nearly full-span features give the same direction. This does not support an RNA-abundance-increase rationale for sirolimus in this model. | Conditional |
No matching records. Change or reset the filters.
Measurements, with their context.
Separate assays, separate scales. No clinical probabilities or efficacy scores.
MVA-NMD-003
One drug, three different reporter responses
NMD / non-NMD normalized RNA ratio · zero-based scale
Ptbp2HnrnplTra2b
Dashed line: reference ratio 1.
View exact chart values
| Condition | Ptbp2 | Hnrnpl | Tra2b | Context |
|---|---|---|---|---|
| Montelukast sodium | 1.3153820850896314 | 3.406494677098417 | 1.7654686351224798 | Published permutation q = 0.02115; not a BUBR1 endpoint. |
| Artesunate | 1.718112969537028 | 1.0078988102818638 | 4.2875477079256585 | Published permutation q = 0.010071; not a BUBR1 endpoint. |
| Cladribine | 2.126124275101715 | 2.0820411294900216 | 1.6377191943091594 | Published permutation q = 0.010071; not a BUBR1 endpoint. |
| Albendazole | 1.5634627380948676 | 1.701726223666598 | 3.2992143735876955 | Published permutation q = 0.015667; not a BUBR1 endpoint. |
| Topotecan hydrochloride | 3.9533398228039465 | 2.3994174963086077 | 0.6542839291174242 | Published permutation q = 0.0 (finite-sampling zero, not zero uncertainty); not a BUBR1 endpoint. |
| Homoharringtonine | 5.0039366160196295 | 24.97590928901173 | 11.672927661494484 | Published permutation q = 0.0 (finite-sampling zero, not zero uncertainty); not a BUBR1 endpoint. |
| Mefloquine hydrochloride | 1.181600971576268 | 1.3075083215610865 | 1.5102164979716604 | Published permutation q = 0.061224; not a BUBR1 endpoint. |
| Nicotinic acid | 1.1859031685137231 | 1.4152893426077033 | 0.9171679519898492 | Published permutation q = 0.08225; not a BUBR1 endpoint. |
MVA-RT-002
Nearby contexts are not the exact variant
Median reporter readthrough (%) · zero-based scale
Median
View exact chart values
| Condition | Median | Context |
|---|---|---|
| Untreated | 0.1745 | 12 nearest measured six-base contexts; exact L737* context absent. |
| Gentamicin | 0.1985 | Same 12 contexts, not BUB1B efficacy or a patient prediction. |
MVA-CTRL-006
Checkpoint competence needs its own control
Median NEBD-to-exit time (minutes) · zero-based scale
Median
View exact chart values
| Condition | Median | Context |
|---|---|---|
| Wild type | 650 | 27 cells; representative replicate, not independent biological replicates. |
| D137R | 60 | 39 cells; representative replicate, not independent biological replicates. |
MVA-RAP-GSE65145-001
BUB1B-spanning RNA feature in normal human fibroblasts
RPKM as deposited; genomic-span RNA feature · zero-based scale
Biological replicate 1Biological replicate 2
View exact chart values
| Condition | Biological replicate 1 | Biological replicate 2 | Context |
|---|---|---|---|
| Proliferative vehicle | 6.227185 | 5.570779 | 10% FBS; DMSO vehicle. |
| Reduced serum | 1.2674783 | 0.9291951 | 0.5% FBS for five days; not matched for cell-cycle distribution. |
| Rapamycin 500 nM | 3.7114706 | 3.198672 | 500 nM for five days; 10% FBS. Mean is 0.585706 times proliferative control. |
Evidence & next validation gates.
Sources describe published experiments. Proposed validation below has not been performed by this project.
MVA-NMD-003704-drug NMD screen: reporter specificityReanalysis
- Evidence class
- COMPUTATIONAL / PATHWAY EVIDENCE
- Model / scope
- Published mouse Neuro-2a AS-NMD isoform summaries; three reporter genes, not BUB1B.
- Finding
- Montelukast has a published permutation q of 0.021150, but the response is reporter-dependent and confounded by non-NMD isoform changes. No hit establishes functional BUBR1 rescue.
- Decision
- Conditional
- Next validation — proposed
- Establish L737* residual function first; any NMD candidate then needs allele-specific RNA, protein synthesis, and separate checkpoint/segregation endpoints.
Limitations and contradictory evidence
- Normalized RNA ratios cannot isolate decay from splicing or isoform regulation.
- Replicate Ct values are unavailable. Published ratio SEM formulas do not provide valid replicate-based uncertainty; no error bars are shown.
- Translation inhibition is counterevidence for homoharringtonine. Statistical significance is not therapeutic suitability.
Sources
- Zhao et al. — molecular profiling of NMD modulators ↗PMID 35024243 · DOI 10.1016/j.omtn.2021.12.003
Mouse Neuro-2a; three AS-NMD isoform reporters · Tables S1/S2/S4; Figures 2–5 and 7
Provenance
- Source artifact
nmd_reanalysis_20261009/results/selected_candidates.csv- Artifact SHA-256
ad7fa784e6caf2f126c9730e4edaf7dc75f0ea2ad54208661de75555ede6ab9c- Method
- Extracted named compounds and ratios from the existing independently reconstructed Table S2/S4 analysis; published q values are not clinical probabilities.
- Verification scope
- Imported an existing analysis artifact and computed its SHA-256. This snapshot build does not rerun the original analysis or validate biological efficacy.
Artifact names and hashes identify the imported analysis package; they are not links to patient data or a claim that unpublished artifacts exist in the public research repository.
MVA-RT-002L737* readthrough: exact versus nearby contextsReanalysis
- Evidence class
- COMPUTATIONAL / PATHWAY EVIDENCE
- Model / scope
- Published 144-nt PTC reporters in human cells; nearest-context subset, not endogenous BUB1B.
- Finding
- No exact L737* measurement or exact six-flanking-base match in the 5,837-PTC table. Twelve nearest contexts have a median paired gentamicin increment of 0.0165 percentage points.
- Decision
- Deprioritized
- Next validation — proposed
- Use the exact L737* context with matched controls and independently test full-length protein, checkpoint response and segregation fidelity.
Limitations and contradictory evidence
- Neighboring-context values are not a prediction for L737* or the child.
- The main predictive analysis excluded gentamicin for weak signal.
- Readthrough does not guarantee native leucine, endogenous transcript survival, or functional BUBR1. Gentamicin plus mefloquine remains an experimental branch only.
Sources
- Toledano et al. — genome-scale PTC readthrough ↗PMID 39174735 · DOI 10.1038/s41588-024-01878-5
Human HEK293T landing-pad reporters; not endogenous BUB1B · Supplementary Table 3; Extended Data Figure 4
- Omachi et al. — COL4A5 readthrough reporters ↗PMID 35243249 · DOI 10.1016/j.isci.2022.103891
HEK293 COL4A5 cDNA reporters; different gene and context · Figure 6: gentamicin plus mefloquine
Provenance
- Source artifact
l737_readthrough_gate_2026-10-09/results/results.json- Artifact SHA-256
8d1f7ca32b05f9e85cedc5d025ea04efee60b581ae4effea8fd389e6f4afbcb9- Method
- Extracted medians from the existing nearest-context reanalysis. The paired median increment is not the difference of group medians.
- Verification scope
- Imported an existing analysis artifact and computed its SHA-256. This snapshot build does not rerun the original analysis or validate biological efficacy.
Artifact names and hashes identify the imported analysis package; they are not links to patient data or a claim that unpublished artifacts exist in the public research repository.
MVA-CTRL-006BUBR1 functional controls: arrest durationReanalysis
- Evidence class
- COMPUTATIONAL / DIRECT-BUBR1
- Model / scope
- Published HeLa BUBR1 depletion/replacement with WT or engineered TPR variants under nocodazole challenge.
- Finding
- Figure 4B source-data medians are 650 minutes for WT and 60 minutes for D137R. A control for checkpoint competence, not a drug-rescue result.
- Decision
- Control only
- Next validation — proposed
- Require independent checkpoint and chromosome-segregation readouts in future equal-expression assays.
Limitations and contradictory evidence
- D137R is not N1002K; no tested drug or exact patient genotype.
- Cells in the displayed representative replicate are not independent biological replicates.
- Mitotic-arrest duration under spindle challenge is not chromosome-segregation fidelity in unperturbed divisions.
Sources
- BUBR1 TPR / MCC–APC/C functional controls ↗PMID 42477419 · DOI 10.1038/s42003-026-10384-9
HeLa endogenous BUBR1 RNAi and RNAi-resistant YFP replacement · Figure 4B; supplementary source-data workbook, Sheet1
Provenance
- Source artifact
bubr1_checkpoint_kinetics_2026-10-09/results.json- Artifact SHA-256
d6be3833115ac368fffe185dfea0e42439e4c80e723b88c626ecb9b874073e6c- Method
- Selected the Figure 4B rows from a completed deterministic source-workbook reanalysis; no kinetic rate or new confidence interval inferred.
- Verification scope
- Imported an existing analysis artifact and computed its SHA-256. This snapshot build does not rerun the original analysis or validate biological efficacy.
Artifact names and hashes identify the imported analysis package; they are not links to patient data or a claim that unpublished artifacts exist in the public research repository.
MVA-N1002K-GATE-002N1002K stabilization: equal-expression gateEvidence audit
- Evidence class
- DIRECT-BUBR1 COMPARATORS / RELATED-VARIANT IN VIVO / TARGETED AUDIT
- Model / scope
- Published human-cell BUBR1 depletion/replacement studies, MVA fibroblast rescue and a mouse L1012P comparator model; none tests N1002K.
- Finding
- No exact-N1002K functional experiment was located. I909T/L1012P checkpoint rescue at matched abundance makes an abundance-limited mechanism plausible, but equally expressed B56-binding mutants still failed chromosome-congression rescue and mouse L1012P showed allelic effects beyond total protein level. Stabilization remains conditional; no drug is promoted.
- Decision
- Conditional
- Next validation — proposed
- At measured matched abundance, compare WT, N1002K and L1012P in one depletion-replacement system; report nocodazole checkpoint arrest separately from chromosome alignment and missegregation per completed division. Wet-lab work is deferred.
Limitations and contradictory evidence
- This targeted audit is not a systematic review; failure to locate an exact-N1002K study does not prove that none exists.
- I909T and L1012P are comparator variants, not substitutes for N1002K.
- B56-binding mutants lie in a different region and demonstrate only that equal abundance is not universally sufficient.
- Mouse L1002P models human L1012P, not human N1002K.
- Checkpoint-only rescue must not be relabeled as chromosome-alignment or segregation rescue.
- The published 6% versus 13% titration is cell-line-specific, not a universal functional or clinical threshold.
Sources
- Suijkerbuijk et al. — molecular causes of BUBR1 dysfunction ↗PMID 20516114 · DOI 10.1158/0008-5472.CAN-09-4319
U2OS/HeLa BUBR1 depletion and replacement · Figures 4D and 5; comparator variants, not N1002K
- Xu et al. — BUBR1 recruits PP2A-B56 for chromosome congression ↗PMID 23789096 · DOI 10.1242/bio.20134051
HeLa BUBR1 depletion/reconstitution and BUB1B-mutated MVA fibroblast rescue · Figures 2 and 4; equally expressed B56-binding mutants are functional counterexamples
- Sieben et al. — allelic effects in MVA progeria syndrome ↗PMID 31738183 · DOI 10.1172/JCI126863
Mouse BubR1 L1002P ortholog model of human L1012P with hypomorphic or truncating alleles · Abstract and Figures 1-3; related variant, not N1002K
Provenance
- Source artifact
n1002k_equal_expression_audit_2026-10-10/audit.json- Artifact SHA-256
30a149370a4bd736fbe39618c2e0ed549008066ea004bd24a5e699308657c73d- Method
- Targeted exact-variant search plus primary-study comparison of abundance-matched BUBR1 rescue and counterexamples; no biological simulation was executed.
- Verification scope
- Primary-source claims were checked against PMID 20516114 Figures 2D/3/4D, PMID 23789096 Figures 2/4 and PMID 31738183 Figures 1-3. Artifact bytes were hashed directly. No independent human scientific review or biological efficacy validation is claimed.
Artifact names and hashes identify the imported analysis package; they are not links to patient data or a claim that unpublished artifacts exist in the public research repository.
MVA-L737-NMD-FIRST-001L737* NMD-first: test truncated-protein functionEvidence audit
- Evidence class
- DIRECT-BUBR1 COMPARATOR / OUR INFERENCE
- Model / scope
- Published 731X/753X BUBR1 constructs and a proposed exact-L737* validation design.
- Finding
- A related 731X construct motivates testing whether L737* retains useful function when NMD is bypassed. The exact L737* experiment has not been performed.
- Decision
- Conditional
- Next validation — proposed
- Test WT, clean L737*, 731X and 753X at matched abundance; advance NMD modulation only if the exact truncation retains useful function.
Limitations and contradictory evidence
- 731X is a frameshift product with about 715 native residues, not a clean L737* surrogate.
- NMD sensitivity is predicted, not measured for this child.
- More nonsense transcript can increase nonfunctional protein; NMD inhibition does not itself remove a stop codon.
Sources
- Suijkerbuijk et al. — molecular causes of BUBR1 dysfunction ↗PMID 20516114 · DOI 10.1158/0008-5472.CAN-09-4319
U2OS/HeLa BUBR1 depletion and replacement · Figures 4D and 5; comparator variants, not N1002K
Provenance
- Source artifact
l737_nmd_first_gate_2026-10-09/experiment.json- Artifact SHA-256
68b890531733aec6fc1f4a6cace8ce8e85bdcccf63bb61900688b831aa383a69- Method
- Imported an evidence-supported proposed design. Completed refers to the audit, not execution of the wet-lab validation.
- Verification scope
- Imported an existing analysis artifact and computed its SHA-256. This snapshot build does not rerun the original analysis or validate biological efficacy.
Artifact names and hashes identify the imported analysis package; they are not links to patient data or a claim that unpublished artifacts exist in the public research repository.
MVA-ARI-007Arimoclomol: target engagement and exposure gapEvidence audit
- Evidence class
- PATHWAY EVIDENCE / COMPUTATIONAL / OUR INFERENCE
- Model / scope
- Human-cell heat-shock studies and a nominal molar conversion of FDA-label serum concentrations.
- Finding
- No direct BUBR1 or N1002K rescue evidence was located in the audit. Heat-shock responses are context-dependent; arimoclomol is not promoted to lead candidate.
- Decision
- Deprioritized
- Next validation — proposed
- Require target engagement at relevant exposure plus exact-variant functional rescue before reconsidering prioritization.
Limitations and contradictory evidence
- Published human-cell systems report both conditional induction and negative responses.
- Total serum exposure and nominal cell-culture concentration are not pharmacokinetically interchangeable.
- Approval for a different disease does not establish MVA efficacy.
Sources
- Deane and Brown — conditional heat-shock co-induction ↗PMID 27273088 · DOI 10.1007/s12192-016-0708-2
Differentiated human SH-SY5Y cells; arimoclomol plus celastrol · Figures 2–5; not a BUBR1 study
- Steurer et al. — HSF1 reporter evaluation ↗PMID 35218516 · DOI 10.1007/s12033-022-00467-3
Human HSF-responsive reporter and endogenous target systems · Table 1 and Figure 4: negative arimoclomol response in this system
- FDA — MIPLYFFA prescribing information (2024) ↗
Regulatory label; human serum pharmacokinetics, different indication · Clinical pharmacology; historical label used by the 2026-10-09 audit
Provenance
- Source artifact
arimoclomol_n1002k_gate_2026-10-09/results.json- Artifact SHA-256
6ce51099f036cd0082358c22176f9f58e3065fc545a32781d40014c020233764- Method
- Imported the existing nominal concentration conversion; no tissue distribution, free-drug exposure or rescue model was calculated.
- Verification scope
- Imported an existing analysis artifact and computed its SHA-256. This snapshot build does not rerun the original analysis or validate biological efficacy.
Artifact names and hashes identify the imported analysis package; they are not links to patient data or a claim that unpublished artifacts exist in the public research repository.
MVA-FUNC-005Cytostasis confounding: denominator unit testSoftware test
- Evidence class
- SYNTHETIC SOFTWARE FIXTURE
- Model / scope
- Invented test counts used only to test code; not an experiment or a biological simulation.
- Finding
- The synthetic fixture flags fewer raw error-positive divisions without a lower error rate per completed division. It does not measure sirolimus activity.
- Decision
- Control only
- Next validation — proposed
- Reuse the denominator checks with prospectively designed real measurements when wet-lab work becomes available.
Limitations and contradictory evidence
- All fixture counts are synthetic.
- No drug response, BUBR1 rescue, or patient outcome can be inferred from a passing software test.
Sources
- Project-generated synthetic software fixture. Not a published biological result.
Provenance
- Source artifact
mitosis_adjudicator_2026-10-08/examples/cytostasis_results.json- Artifact SHA-256
dfb5fa29aa719fc59d206cf34a8d33b1ce59bda41d5665f685a219c6770a753b- Method
- Imported the labeled fixture output, checked the synthetic flag, and excluded synthetic quantities from the scientific charts.
- Verification scope
- Imported an existing analysis artifact and computed its SHA-256. This snapshot build does not rerun the original analysis or validate biological efficacy.
Artifact names and hashes identify the imported analysis package; they are not links to patient data or a claim that unpublished artifacts exist in the public research repository.
MVA-RAP-GSE65145-001Rapamycin in human fibroblasts: BUB1B locus RNA signalReanalysis
- Evidence class
- Public human perturbation data; descriptive locus-level RNA reanalysis, not functional rescue.
- Model / scope
- Normal 2DD fibroblasts: proliferative vehicle, 0.5% serum, or 500 nM rapamycin for five days; two biological replicates per condition.
- Finding
- The primary BUB1B-spanning feature has a Rap/Pro mean ratio of 0.585706 (41.43% lower); reduced-serum/Pro is 0.186191. Three nearly full-span features give the same direction. This does not support an RNA-abundance-increase rationale for sirolimus in this model.
- Decision
- Conditional
- Next validation — proposed
- Keep sirolimus conditional. Proposed: compare full-length BUBR1 protein and segregation errors per completed division after transient exposure and washout at matched cell-cycle distribution. Wet-lab work is deferred.
Limitations and contradictory evidence
- Unstranded genomic-span features, not direct BUBR1 protein measurements or validated mature transcript isoform estimates.
- Three overlapping features are sensitivity views of the same locus, not independent biological replicates.
- Two biological replicates per condition from one normal cell line; no MVA genotype and no donor-level replication.
- Serum reduction is a useful control but is not matched to rapamycin for cell-cycle distribution; no causal proliferation adjustment is claimed.
- Continuous 500 nM/five-day treatment; no washout, transient exposure, aneuploidy, mitotic checkpoint or rescue endpoint.
- Published normalized RPKM values are used as deposited without re-alignment, count-based differential expression, an added pseudocount, p-values or rescue probabilities.
- 149122 rows were observed in each deposited track; the paper states 149135 probes. The 13-row difference is not resolved here.
- This does not establish that BUBR1 protein falls, rule out post-transcriptional effects, or predict clinical harm or benefit.
Sources
- Gillespie et al. — rapamycin and normal human fibroblast transcription ↗PMID 26652669 · DOI 10.1080/19491034.2015.1128610
2DD normal human foreskin fibroblasts; two biological replicates per condition · GSE65145 processed tracks; Figures 1 and 3; Figure S2; Cell culture and RNA-seq methods
Provenance
- Source artifact
MVA_GSE65145_BUB1B_2026-10-10.zip/gse65145_bub1b_results.json- Artifact SHA-256
e71199b4f23c6315f02febfd18aac8b9bcd7e7f8bb05a7a7b624ad874d54d3b6- Method
- Parse six deposited hg19 RPKM tracks; require >=99% reciprocal overlap with RefSeq NM_001211.6; select minimum boundary-distance feature and calculate ratios of arithmetic means. Check all three qualifying features.
- Verification scope
- All six coordinate arrays match; each has 149122 rows. Eighteen rows overlap the anchor and three qualify. GEO sample titles and artifact hashes verified. Complete reproducible package saved separately; no human scientific review or new biological experiment is claimed.
Artifact names and hashes identify the imported analysis package; they are not links to patient data or a claim that unpublished artifacts exist in the public research repository.
THE QUESTION
Rare Disease, Real Kid
Behind the data is a real child with mosaic variegated aneuploidy (MVA), a rare disorder of chromosome segregation that combines a predisposition to childhood cancer with severe growth restriction.
Given the patient's genome, which variants explain the disease?
THE HYPOTHESIS
Two BUB1B variants, a causal-pair hypothesis
BUB1B encodes BubR1, a core component of the spindle assembly checkpoint. The pair includes a premature stop and a missense substitution in the C-terminal region. Molecular phase, residual function and pharmacologic rescue remain unresolved.
Allele 1 · truncating
c.2210T>Gp.Leu737Ter
Introduces a stop codon at residue 737 before the C-terminal kinase/pseudokinase region. The transcript is predicted to be NMD-sensitive; its abundance and any residual truncated-protein function have not been measured for this case. ClinVar annotations associate this variant with MVA syndrome type 1.
Allele 2 · missense
c.3006T>Gp.Asn1002Lys
Ultra-rare: observed once among ~1.46 million alleles in gnomAD v4.1, with no homozygotes. Asparagine to lysine at residue 1002, inside the BubR1 kinase domain.
SIFT 0.000PolyPhen 0.997CADD 26.4
WHAT WE DID NOT PROVE
The limits, in writing
The VCF carried no phase information and there was no parental segregation data. We could not independently show that the two alleles are in trans, one inherited from each parent.
We left it in the report as an open limitation, together with the validation path that would resolve it. A genomic result that does not state what it did not prove is not a result; it is a claim.
INSIDE THE CELL
From allele to phenotype
- 01Candidate causal pair
p.Leu737Ter and p.Asn1002Lys motivate a reduced-functional-BubR1 hypothesis; residual function is unmeasured.
- 02Weak checkpoint
The spindle checkpoint no longer holds anaphase until every chromosome is attached.
- 03Lagging chromosome
Division goes ahead anyway and a chromosome is left behind.
- 04Unequal daughter cells
Daughters inherit n+1 and n−1. Repeated through development, this yields a mosaic body.
This simplified disease model motivates research into future segregation errors. It is not evidence that an intervention can reverse established aneuploidy, developmental abnormalities or cancer.
THE FUNNEL
From 5 million variants to one pair
-
5,012,204
variants in the genome
bcftools view -H "${VCF}" | wc -l -
4,740,790
pass the caller filter
bcftools view -f PASS -
14
in the BUB1B locus (13 PASS)
bcftools view -r 15:40161069-40221123 -
7
PASS heterozygous
bcftools view -i 'GT="het"' -
2
exonic on NM_001211.6 (MANE)
VEP --offline, MANE transcript -
1 pair
the hypothesis presented
ClinVar · gnomAD · SIFT · PolyPhen · CADD · domain
Driven by the disease mechanism rather than brute force: prioritise the gene, extract and QC the locus, map to the MANE transcript, and require every candidate to survive converging evidence.
REPRODUCIBLE BY DESIGN
One Makefile, one container
Every target runs inside the container. Whoever clones the repository runs exactly what we ran.
make build # Ubuntu 24.04 ARM64 + bcftools, samtools, bedtools make check # every tool answers before touching data make qc # quality control of the full VCF make bub1b # targeted extraction of the locus make bub1b-phase # phase review of the candidates # Functional annotation, offline against a local cache vep --cache --offline --dir_cache /vep_data \ --species homo_sapiens --assembly GRCh38 \ --symbol --canonical --mane --sift b --polyphen b \ --af --af_gnomade --af_gnomadg --max_af --clin_sig # Last step: audit before the repository goes public bash scripts/prepublish_audit.sh
PRIVACY BY DESIGN
Anyone can reproduce the method
Runs on our machine
The whole analysis ran locally in a reproducible container. --offline guarantees no patient coordinate leaves the machine to be resolved by an external service.
Method, not data
The public repository holds the Dockerfile, scripts, pipeline and documentation, and no patient genomic data: no raw VCFs, genotype-level tables or downloaded references.
A script, not a checklist
An audit script runs before the repository is made public. Checking by hand is not a control; a script that fails is.
Acknowledgement
This work was made possible through the Hackathon, organized by Sage Bionetworks in partnership with the MVA Society, Hugging Face, and BEACON (The Benchmarking, Evaluation, and Assessment Consortium for Science), with prize sponsorship from AWS and Anthropic. We are deeply grateful to the child and their family who generously contributed their data and their story to advance research into this rare disease. We acknowledge their trust in making this Hackathon possible.