Separate 5mC and 5hmC readouts improve age-related disease detection
Primarybiomodal's core mechanistic claim is that 5-methylcytosine and 5-hydroxymethylcytosine carry distinct, biologically meaningful information about genome regulation, cell state, and disease. Because traditional methylation approaches can conflate these cytosine modifications, they may obscure disease-relevant epigenetic signals. By measuring 5mC and 5hmC separately alongside the four canonical DNA bases, 6-base sequencing should reveal regulatory changes that better distinguish age-related disease states from healthy controls. A testable prediction is that classifiers using separate 5mC and 5hmC features will outperform classifiers using only aggregate modified cytosine or single-mark features for diseases where epigenetic dysregulation is involved. In colorectal cancer cfDNA, this is supported by higher diagnostic performance when 5mC and 5hmC are combined versus conflated modified-C features.
Popperian evaluation
The premise is credible: 5mC and 5hmC are different cytosine marks, traditional methylation workflows can collapse them into one modified-cytosine signal, and 6-base sequencing can read them separately. The strongest biological point is that colorectal cancer regions showed different behavior for the two marks, including early 5hmC increases and later 5mC decreases. That makes conflation a real measurement problem, not a cosmetic one. The weak spot is generality: the evidence is strongest in colorectal cancer cfDNA, while the broader age-related disease claim still needs disease-by-disease testing.
Supporting evidence: The 2026 Epigenomics review says 6-base sequencing can detect A, C, T, G, 5mC, and 5hmC in one workflow and argues that the two cytosine marks carry different regulatory information.; The colorectal cancer cfDNA study reported that combined 5mC and 5hmC features reached AUC 0.95 versus AUC 0.66 for conflated modified cytosine.; Colorectal cancer differentially methylated regions showed opposing or stage-dependent 5hmC and 5mC patterns, including increased 5hmC in stage I and decreased 5mC in stage IV in many regions.
Counter evidence: The disease-detection evidence is concentrated in colorectal cancer cfDNA, so the age-related disease extension is still partly an assumption.; 5hmC alone can discriminate colorectal cancer from controls, which means separate dual-mark measurement may help most in some settings and add less in others.
The theory explains the main observation well: if 5mC and 5hmC move differently during disease, collapsing them should erase signal, and separating them should improve classifiers. The AUC gap in colorectal cancer cfDNA, 0.95 versus 0.66, fits that logic cleanly. Still, alternative explanations remain possible. Better performance could partly reflect richer feature count, model fitting, cancer-specific biology, cohort structure, or the choice of DMRs from stage IV tissue. The theory explains the pattern, but it has not yet beaten those alternatives across enough diseases.
Supporting evidence: Combined 5mC and 5hmC models outperformed conflated modified-cytosine models in colorectal cancer cfDNA.; Opposing 5hmC and 5mC behavior across colorectal cancer stages gives a mechanism for why aggregation can hide signal.; A separate 5hmC cfDNA study in more than 2,000 blood samples supports hydroxymethylation as a real cancer signal, including early colorectal cancer.
Counter evidence: The reported classifier gain could come from added feature richness rather than the specific biological interpretation of separate 5mC and 5hmC signals.; The current evidence does not show the same performance gain across multiple age-related diseases with matched controls and external validation.
This is strongly falsifiable. The prediction is concrete: in diseases with epigenetic dysregulation, classifiers using separate 5mC and 5hmC features should outperform aggregate modified-C models and, when both marks carry signal, single-mark models. A fair test can predefine cohorts, features, model classes, train-test splits, and endpoints such as AUC, sensitivity at fixed specificity, or calibration. If dual-mark models fail to improve on aggregate or single-mark models in external cohorts, the central detection claim takes a direct hit.
Supporting evidence: The theory names the comparison classes: separate 5mC plus 5hmC, aggregate modified cytosine, 5mC-only, and 5hmC-only.; The colorectal cancer cfDNA result gives a measurable benchmark, AUC 0.95 for combined marks versus 0.66 for modified C.; The prediction can be tested prospectively in independent disease cohorts with standard classifier metrics.
Counter evidence: The phrase 'diseases where epigenetic dysregulation is involved' needs pre-specified inclusion criteria, otherwise failed tests could be dismissed after the fact.; Classifier comparisons can become muddy if feature counts, sequencing depth, preprocessing, or model complexity differ across arms.
Reasoning tree
Public endorsements
The dossier identifies Peter Fromen as biomodal's CEO, but it does not provide any public quote, talk, article, or statement from him addressing the theory that separate 5mC and 5hmC readouts improve age-related disease detection. On this evidence, he stays silent on the claim itself.
The only evidence provided is biomodal's team page listing Fiona Stewart on the leadership team. It does not contain a statement from her about separating 5mC and 5hmC readouts, so there is no public endorsement, mention, or contradiction of the theory in this record.
Fromen appears to speak publicly about biomodal's "true 6-base genome" being powerful for biomarker discovery, which points in the same direction as the theory that separating modified cytosine signals adds diagnostic value. But the evidence here does not show him explicitly saying that separate 5mC and 5hmC readouts improve age-related disease detection, so this is a mention rather than a clear public endorsement.
Robert Osborne appears publicly in biomodal-linked materials that directly advance this claim. A podcast featuring him is titled "Separating Epigenetic Signals Improves Early Cancer Detection," and a 2026 biomodal event summary says Osborne presented science using biomodal's 5- and 6-base sequencing technology for earlier biomarker discovery and disease mechanism analysis. That is an endorsement of the core idea that separating epigenetic signals improves disease detection, not silence or contradiction.
Evidence publication IDs: cf295889-6420-4541-bf0b-a3fe88b38db6, 6f91e672-216f-48de-b246-9fb5ded0e9ee
The record shows Balasubramanian publicly backing the company and appearing as its founder, but it does not show him explicitly stating the theory that separate 5mC and 5hmC readouts improve age-related disease detection. The financing quote supports company-level endorsement, not this mechanistic claim, and the event listing gives a title without the underlying statement.