REFERENCE EVIDENCE
Established
Definitive autosomal-dominant gene-disease validity, sufficient evidence for haploinsufficiency, published human phenotypes, and substantial structural and model-system biology.
RESEARCH PROGRAM
A research program built around testable phenotypes.
The 61-person retrospective evidence corpus identifies candidate natural-history variables, molecular subgroups, and transition or recovery questions that require prospective measurement. Published human, cellular, and model-system studies define the experimental context for investigator-led work.
STRUCTURED RETROSPECTIVE EVIDENCE
Mapped
Sixty-one connected people across 19 countries anchor the roster; 51 have mapped phenotype evidence and 15 have deep age-ordered packets. Molecular architecture, age, sex, relatedness, source depth, co-findings, 15 clinical lanes, and chronology are used only where their denominators support the question.
PROSPECTIVE PROGRAM
To test
Natural history, subgroup effects, state physiology, extra-neural biology, biomarkers, prognosis, and intervention response under investigator-led protocols.
PROPOSED CAUSAL STUDY ARCHITECTURE
A credible ASH1L mechanism must survive allele × cell × state × time.
Human genetics establishes dosage sensitivity, and experimental studies show that ASH1L effects can depend on locus, lineage, and developmental state. The next decisive study should preserve those dimensions in one design rather than testing a downstream pathway in isolation.
THE EXPERIMENT THAT CAN SEPARATE COMPETING MODELS
One allele-aware design. Four controlled dimensions. Six timepoints. One rescue gate.
Direct experimental anchorProposed ASH1L study
Preserve the molecular question.
Patient-derived allele, CRISPR-corrected isogenic control, engineered dosage series, matched reference, and ASH1L rescue.
Compare identities—not “brain” versus “body.”
Direct neural anchor: progenitor, immature and mature glutamatergic, and mature GABAergic states. Priority gaps: astrocyte and barrier epithelium.
Define what changes the system.
Use a controlled maturation step or tissue-appropriate challenge with matched dose, duration, culture state, and recovery conditions.
Sample the return—not only the peak.
Repeated measurement must distinguish lower baseline output, delayed transition, excessive activation, and slow recovery.
SAME CELLS · SAME ASSAYS · ORDERED TIME
Every point is compared across patient allele, isogenic correction, dosage series, and rescue.
- t0
Baseline
Residual dosage and stable function
- t1
Perturb
Defined maturation step or challenge
- t2
Peak
Magnitude and pathway engaged
- t3
Close
Transcriptional or functional shutoff
- t4
Recover
Return toward the starting state
- t5
Re-challenge
Reset, memory, and recurrence threshold
Is functional ASH1L actually reduced?
Allele-specific RNA and long-read transcript analysis, transcript stability, quantitative full-length protein, localization, catalytic function, and complex assembly.
Where does chromatin regulation change?
CUT&RUN or CUT&Tag for H3K36me1/2, H3K27me3, H3K4me3, and ASH1L occupancy; pair with ATAC-seq at predefined loci and report bulk marks separately.
Which programs mistime?
Time-resolved RNA-seq, splicing, and pathway activity to separate delayed induction, altered amplitude, delayed shutoff, and incomplete recovery.
Does the molecular difference matter to the cell?
Proliferation and differentiation, multielectrode-array or patch-clamp physiology, transmitter and ion handling, epithelial resistance, transport, repair, or metabolic throughput.
Does the same measurable pattern exist in people?
Repeat a defined clinical or physiological measure at baseline, during a prospectively specified load window, and through recovery.
RESCUE & REPLICATION GATE
Advance a target only when the levels converge.
- Correction or rescue normalizes the readoutSupports ASH1L dependence
- The effect is restricted to a cell stateDefines lineage or maturation vulnerability
- The peak is similar but recovery divergesSupports a specific resolution hypothesis
- The difference does not rescueElevates background, co-finding, or assay explanations
PRE-SPECIFIED FALSIFIERS
A strong framework states what would make it wrong.
- 01
No allele-dependent reduction or alteration of functional ASH1L is measurable.
- 02
Chromatin differences do not localize to the same regulatory loci or precede the transcriptional change.
- 03
The cell-state or recovery phenotype does not normalize after genetic correction or ASH1L rescue.
The proposed chain should be revised or rejected if these tests fail; a negative result is mechanistically informative.
Direct ASH1L knockout effects varied across modeled neural cell states; complete knockout is not a patient-allele dosage series.
Fernandez Garcia et al. ↗ASH1L chromatin-engagement domains were resolved, and Ash1l depletion altered the timing of selected differentiation-marker expression in mouse ES models.
Vann et al. ↗In mouse mesenchymal stem cells, ASH1L-associated H3K36me2 was highly locus-specific; bulk abundance alone would have missed that signal.
Shipman et al. ↗CURRENT SCIENTIFIC CONTEXT
Four evidence streams now define the next study.
Human genetics, functional genomics, mechanistic models, and longitudinal clinical data each answer a different part of the ASH1L problem.
Human clinical genetics
Published cohorts establish a heterogeneous ASH1L-related neurodevelopmental phenotype and a primary loss-of-function mechanism. Variant class and protein position have not yet yielded a simple, validated severity ladder.
Human studies →Yale functional genomics
The 2026 Nature Neuroscience study knocked out ASH1L across four neural cell states within a 23-gene screen. ASH1L loss contributed to a mitochondrial membrane-potential signal in the pooled screen, while arrayed ASH1L knockout did not significantly change Seahorse respiratory parameters. It is direct gene-loss evidence—not a patient-allele dosage assay or a mitochondrial diagnosis.
Read the mitochondrial evidence boundary →He Lab models and active questions
Peer-reviewed Ash1l mouse studies support developmental, transcriptional, circuit, sleep, seizure-threshold, and excitation/inhibition questions. The NIH-funded program separately examines excitatory-neuron, inhibitory-neuron, and astrocyte contributions; those lineage-specific studies are active program aims, not completed peer-reviewed human evidence.
He Lab research ↗Longitudinal cohort architecture
The connected dataset adds what cross-sectional databases rarely capture: age-ordered change, state context, recovery, adult trajectories, source asymmetry, and multisystem co-occurrence within the same person.
Longitudinal phenotype →2026 estradiol study
This preprint reports ASH1L responses in human induced neurons and larval zebrafish. It supports replication, dose–response, timing, and mechanism studies before any human interventional interpretation.
Review the preprint record ↗NEXT RESEARCH PHASE
Eight open questions now define the path forward.
Natural history, variant mechanism, age and sex, state physiology, multisystem biology, biomarkers, intervention windows, and meaningful trial outcomes each require a specific study design.
COLLABORATION MODEL
Clinical depth, mechanistic expertise, and structured participation.
The program is designed for investigator-led protocol development with appropriate ethics review, consent, data-sharing, authorship, and return-of-results structures.
FamiliesLongitudinal history, function, record access, and research priorities.
CliniciansPhenotype validation, examination, differential interpretation, and clinical safety.
ResearchersProtocol design, assays, statistics, stewardship, and reproducibility.