Evidence-qualified scientific claims: every statement below carries its supporting evidence and scope.
Updated October 16, 2026
Bacteroides methyltransferase BF2170 methylates bezafibrate, raising epithelial permeability ~20-fold in Caco-2 assays and plasma drug exposure at 7–9 h in gnotobiotic mice — but only 15 of 170 tested carboxyl drugs were methylated and generalization to complex human microbiomes is not yet shown.
Microfluidics, genetics, and in vivo imaging show hook flexibility (FlgE) causally enables wrapped-flagellar translocation through micrometer-scale constrictions across 13 Burkholderia sensu lato species, though evidence is limited to rectangular channels and one symbiont model.
Electron microscopy and whole-genome sequencing reveal envelope bridges (~63 nm) mediating chromosomal transfer up to 1.1 Mbp in one tested strain pair — a genuinely novel mechanism whose formation machinery remains unknown.
Congenic cured/re-infected mice show the dsRNA mycovirus AfuPmV-1M enhances lung virulence, and ribavirin lowered viral load and improved survival — strong within one isolate and mouse models, untested across clinical isolates and human disease.
Each mechanistic link (fiber loss → SCFA-producer decline → barrier/immune homeostasis → inflammation) has separate support in animals and humans, but no direct study tests the full chain with ultra-processed foods themselves.
The VDAART cohort precedent (33–55% mediated effects) shows microbiome mediation of perinatal exposures is detectable at cohort scale, but only with ≥400 maternal samples and shotgun-metagenomics resolution — no direct dye-biomarker cohort exists yet.
Enformer predictions across 34 hominin/ape haplotypes identified 37,592 lineage-specific regulatory elements with one causally validated base-pair mechanism (SRF motif at GNB5), but only a small, non-random validation subset was tested.
Repurposing single-nucleus ATAC-seq on 3,842 haploid mouse genomes yielded >25,000 crossovers and confirmed elevated rates in Fancm-deficient mice, with per-cell genetic-distance recovery somewhat lower than prior methods.
Four artificial proteorhodopsins absorbing at ~410 nm increased C. necator growth in a preprint — but proton pumping was inferred, not directly measured, and the design pipeline relies on proprietary platforms.
Equilibrium chemistry explains why cytosine and thymine are absent from meteorites while A, G, and U persist, though the model overproduces bases by 3–4 orders of magnitude and its H₂O₂ selectivity claim is untested.
A preprint maps model training onto drift, immigration, blending, and recombination with verified closed forms and pre-registered falsifiers, but key mechanisms were tested only in small simulations and two LLM families.
A factorial design ( bacterium strain × inhibitor treatment, n≥8/group) with LC-MS, telomerase, and RNA-seq endpoints is adequately powered; the inhibitor-occupancy-without-transcription-change criterion cleanly separates pharmacological from biological failure.
A North Asian mitochondrial lineage (C4a2c1) persists in Kurumba without detectable autosomal Northeast Asian affinity — plausible but asserted rather than simulated, with per-cluster samples under five.
In a >20-year exclusion experiment, drought-primed trees lost transpiration ~2× slower during the 2023 El Niño and recovered fully within 143 days — compelling but based on a single plot pair, with resilience robustness in only 71.5% of bootstrap resamples.
RNAi knockdown of a nicotinic receptor subunit (chrnal-E) increased sleep in Cassiopea, the first functional genetic sleep evidence in a cnidarian — but small samples, no rescue experiment, and stock-species uncertainty temper the claim (preprint).
Multi-level preprint evidence (human postmortem, mouse behavior, electrophysiology) supports sex-opposite regulation of one extracellular-matrix protease in nucleus accumbens, though human causality and BCAN rescue remain untested.
Perceived isolation predicted decline in 823 older adults over 65 months, yet autopsy data showed no link to neuropathology — a reserve-versus-disease confound still unresolved in the loneliness-cognition field.
A VAE decoder generated intuitively interpretable scene hypotheses for 11 fMRI-defined navigation regions (preprint), but validation is largely qualitative, n=6 includes two authors, and reconstruction quality was not quantitatively scored.
Source-space MEG in 12 expert monks showed that aperiodic correction reverses the gamma-increase narrative (preprint) — methodologically careful but N=10 analyzed, no control group, and the E/I interpretation remains speculative.
Two small naturalistic studies (n=8 basketball, n=21 soccer) found vicarious competition shifts testosterone, overturning normal diurnal decline in winning fans — foundational but underpowered, all-male, and without diurnal-timing controls.
A BJPsych Open conceptual paper proposes a chatbot-vulnerability × sycophancy × hallucination loop for chatbot-associated psychosis — clinically actionable suggestions, but the mechanism is an untested hypothesis resting on case reports.
Only one high-quality MMR–ASD study exists in the supplied evidence (OR 0.86, 95% CI 0.68–1.09, UK case-control), so no pooled estimate or heterogeneity analysis is possible without inflating the evidence base.
Rutter's 2005 review concludes true ASD incidence is likely 30–60 per 10,000 (up from 4), with the rise largely from ascertainment and diagnostic broadening and no epidemiological support for MMR or thimerosal causation.
A review of 50 observational studies finds roughly four in five report positive depression associations, with anxiety evidence weaker and sometimes U-shaped — narrative-only synthesis with unvalidated self-reported exposure measures.
A meta-analysis of 23 prospective cohorts (181,006 participants) finds ~29% higher CHD and ~32% higher stroke risk, but funnel asymmetry suggests the true effect may sit closer to 18–19% than the headline 30%.
A Navigation Guide systematic review rates prenatal pyrethroid exposure as having "sufficient evidence" of adverse neurodevelopment at moderate quality, with the strongest mechanistic support via thyroid hormone disruption — though spot-urine biomarkers and two cohort reports of beneficial associations complicate interpretation.
A spatial cross-sectional study (N=21,851) links PM2.5 to higher sperm DNA fragmentation (estimate 0.73, p<0.0001), with age (+14.4 pp at ≥50) dwarfing the pollution effect — non-causal inference warranted given cross-sectional design and unmeasured smoking/BMI.
Cohorts show modestly lower mortality at 2–3 cups/day while acute caffeine raises blood pressure ~8/6 mm Hg in hypertensive individuals — the contradiction largely reflects acute-vs-habitual time scales, so hypertension-prone people may still carry a daily BP burden.
~150 min/week of moderate activity prevents weight gain while 250–300 min/week supports clinically significant loss and liver-histology improvement in NAFLD — the minimum and optimal doses differ roughly twofold.
A CDC-authored review synthesizes evidence that influenza A viruses can reach mammalian gut tissue and that ingesting contaminated raw milk can establish infection, while judging human risk from consumption low but theoretically possible.
In 1,083 BASE-II adults followed 7.4 years, Allostatic Load and DunedinPACE predicted outcomes most consistently and improved incident-disease models by up to 24 percentage points — though ALI's win partly reflects construct overlap with the outcomes.
A NEJM clinical-practice review supports screening, psychological therapies, and lactation-compatible SSRIs, with antenatal depression the strongest predictor — but the key prevention statistic (0% vs 92%) rests on one small observational study (n=78) that demands replication.
A narrative review competently synthesizes helminth therapy, maggot debridement, and leech therapy mechanisms, but relies almost entirely on preclinical data, and zero recombinant parasite-derived immunomodulators have reached human trials.
Cross-species evidence (zebrafish, mice, C2C12 cells) shows globin digest drives hypertrophy, but no isonitrogenous protein control, no blinding, small samples, and corporate conflicts of interest limit the claimed "peptide relay" mechanism.
A stop-gain variant appeared in 3/101 Polish top-elite athletes and 0/1,899 comparators, but Bonferroni-adjusted p=0.069, no carrier physiology, and self-acknowledged chance make the beneficial-variant hypothesis exploratory.
UK Biobank data (n=190,275) show measured omega-3 levels are protective (aHR 0.86) while genetically predicted levels raise AMI risk (OR 1.11) — implying omega-3 levels partly mark broader metabolic health rather than uniform causal benefit.
A preregistered meta-analysis of 54 effect sizes (n=16,013) finds a moderate association (r≈0.255) between compassion for others and well-being, weakest for negative affect — largely cross-sectional, self-report evidence that cannot establish causation.
Animal and cell studies consistently show inflammatory signaling and oxidative stress, and microplastics amplified cadmium accumulation in zebrafish liver by 46–184% — but the causal human link remains nearly empty, with dose–response translation the field's biggest gap.
Mechanistically plausible animal and in vitro evidence suggests particle size switches harm mode (microparticles: immune/dysbiosis; nanoparticles: apoptosis/barrier damage), but no human causal data exist and the underlying review's full text could not be verified.
An unreviewed preprint reports sodium benzoate increases fat and shortens lifespan via SKN-1/Nrf2 in worms at low doses, but non-monotonic dose effects, missing data, and absent osmolarity controls leave the finding hypothesis-generating only.
No published N-body grid directly reports Earth-analogue survival fractions versus binary separation and eccentricity — the number must come from new simulations, with periastron (not separation) expected to govern the survival collapse.
Know what changed, what holds up, and what remains uncertain. Every Friday. No ads.
Explore by Goal
Ask a Follow-Up
Trending