Browse by Section

Gut bacteria synthesize drug prodrugs in the intestine

Bacteroidota methyltransferase BF2170 converts bezafibrate into a methyl ester in the gut, raising epithelial permeability ~20-fold in cell assays and increasing plasma drug levels in mice — but demonstrated only in mono-colonization models, not complex human microbiomes.

Biofilm plasmid transfer varies 1,500-fold between plasmid types

Clinical IncX1 and IncU plasmids span a 1,500-fold conjugation-frequency range in biofilms, consistent with two-regime spatial-structure models — but whether the bottleneck is mating-pair formation, maintenance cost, or spatial access is unresolved.

Probiotic plasmid blocks acquisition of resistance plasmids

An engineered conjugative plasmid carrying entry-exclusion genes blocks >99.9% of in vitro resistance-plasmid uptake and prevents gut colonization in mice — small groups (n=3), no statistical testing, and untested in complex human microbiota.

Cell-envelope conduits move megabase DNA between bacteria

Acinetobacter baumannii forms continuous envelope bridges (~63 nm) containing DNA-like filaments that enable chromosomal exchange up to 1.1 Mbp (~27% of the genome) — the machinery is unidentified and only one strain pair has been tested functionally.

Sleep rewrites hippocampal maps toward prior knowledge

Post-encoding sleep in mice reinstated familiar CA1 place maps in a novel environment (correlation 0.32 vs 0.10 after wake), and prefrontal inactivation during sleep abolished this — small samples (4–7 mice) and male-only cohort limit confidence.

Cholinergic sleep control in jellyfish predates the brain

RNAi knockdown of nicotinic receptor subunit chrnal-E increased sleep in the upside-down jellyfish Cassiopea — first functional genetic sleep evidence in a pre-bilaterian, though sample sizes are small and no rescue experiment was performed.

Loss aversion may be noradrenergic, reward-seeking dopaminergic

A dual-neurotransmitter model proposes dopamine drives reward-seeking while amygdala noradrenaline drives loss aversion — testable, but resting on small pharmacological studies and a 2-patient lesion report.

Loneliness predicts cognitive decline independent of brain pathology

Perceived isolation predicted cognitive decline and Alzheimer's risk in 823 older adults over 65 months, yet autopsy showed no link to neuropathology — suggesting effects on cognitive reserve rather than disease progression, with causality untested.

Fans' testosterone rises and falls with their team's fortune

Winning fans showed testosterone increases and losing fans decreases in two naturalistic studies (n=8 basketball, n=21 soccer) — foundational for biosocial theory but tiny samples with no alcohol or diurnal-timing controls.

Bats time echolocation calls to gaps in rhythmic noise

Carollia perspicillata phase-locked calls to low-noise phases of 4-Hz modulated noise in lab and wild, with frontal auditory field inactivation disrupting timing — demonstrated for one species and one masker frequency only.

Deep learning maps human regulatory evolution beyond acceleration scans

Enformer predictions across 34 hominin and ape haplotypes identified 37,592 lineage-specific regulatory elements with validated prediction–observation correlation R = 0.81 — but ~35% of elements show below-average divergence, suggesting acceleration-based discovery has under-sampled causal mutations.

Thermodynamic equilibrium explains missing meteorite nucleobases

Equilibrium chemistry in planetesimal interiors explains why cytosine (deamination, <17,000-yr half-life) and thymine (H₂O₂ oxidation) are absent from meteorites while adenine, guanine, and uracil persist — the H₂O₂ selectivity claim remains experimentally untested.

North Asian maternal lineage persists without autosomal trace in Nilgiri tribes

A rare mtDNA lineage (C4a2c1) in Kurumba tribes dated to ~7 ka shows no detectable autosomal Northeast Asian affinity — a caution that mtDNA-only ancestry claims cannot distinguish ancient migration from founder-lineage bottlenecks; tiny per-cluster samples and no data deposit.

Simulated aging as loss of regenerative goal-directedness

Neural cellular automata suggest aging emerges from loss of morphogenetic goals rather than damage accumulation, and that residual information at lost-organ sites ("memory sites") persists — proof-of-concept only, with no empirical validation or public code.

Codon language model advantages shown to be evaluation artifacts

A leakage-controlled benchmark (CodonBench) shows the reported codon-LM advantage on synonymous variants collapses from +2.3–14.3 to ~+1.6–2.2 percentage points under gene-held-out splits, and a memorization canary (AUC 0.739) beats every neural model.

Population genetics maps multigenerational AI model lineages

A formal framework maps model training on generated data to drift, real-data injection to immigration, weight averaging to blending, and merge failure to reproductive isolation — with closed forms and falsifiers, but simulation-only evidence for key mechanisms and small seed counts.

Optimal-control model of human liquid pouring

3,367 measured pours show fill amount is invariant despite speed pressure and that "headroom" serves as the stopping criterion, captured by a SINDYc + inverse optimal control model — data/code public, but noise parameters are hand-tuned and the sample is narrow.

Nano- and microplastic health risks: mechanism strong, human causation thin

Animal and cell studies consistently show inflammatory signaling and oxidative stress from plastic particles, and co-contaminant amplification (cadmium liver accumulation +46–184% in zebrafish) — but the causal human link remains nearly empty, and the exposure-scenario framework paper carries declared plastics-industry funding.

Microplastics and gut microbiome disruption claims reviewed critically

Gut-dysbiosis and chronic-disease claims are mechanistically plausible from animal evidence but rest on narrative, non-systematic reviews with no quantitative risk thresholds — particle size may switch the harm mode (immune dysbiosis vs barrier apoptosis).

Pyrethroid insecticides and child neurodevelopment: "probably toxic"

A Navigation Guide review of 25 studies rates prenatal pyrethroid exposure as having sufficient evidence of adverse neurodevelopment, with thyroid disruption as the best-supported mechanism — spot-urine biomarkers and single-author study selection temper certainty, and some cohorts reported beneficial associations.

Air pollution (PM2.5) and sperm DNA fragmentation in US men

In 21,851 clinic samples, higher PM2.5 was associated with increased sperm DNA fragmentation (estimate 0.73) — but the age effect (+14.4 pp at ≥50) dwarfs it, and unmeasured smoking/BMI confounding plus cross-sectional design preclude causal inference.

H5N1 influenza and the gastrointestinal route via raw milk

A CDC-authored review finds influenza A viruses, including H5N1, can reach and replicate in mammalian gut tissue and that ingestion of contaminated material can establish infection — judged low but theoretically possible human risk; no validated fecal diagnostics exist yet.

Building 1 kg of muscle costs only ~41 kcal per day

A bottom-up accounting model estimates tissue-specific energy cost at ~41 kcal/day during accretion — a ~10× gap below standard 300–500 kcal surplus prescriptions, which more plausibly serve as insurance against underfeeding; zero primary data, key parameter derived from infants/animals.

Ultra-processed foods, fiber loss, and gut barrier inflammation

The mechanistic chain (substrate loss → butyrate-producer decline → barrier dysfunction → systemic inflammation) is strong at every intermediate link but has never been tested with ultra-processed foods themselves — the attribution rests on observational diet-pattern epidemiology.

Two cups of coffee: resolving the benefits-vs-harms contradiction

Habitual intake of 2–3 cups/day shows modestly lower cardiovascular and all-cause mortality in large cohorts (HR 0.95 in 185,000+), while acute caffeine (200–300 mg) transiently raises blood pressure ~8/6 mmHg in hypertensive individuals — the contradiction is largely a time-scale artifact.

Social media and adolescent depression: association, not causation

A review of 50 studies finds ~4 in 5 report a significant positive association between social media use and adolescent depressive symptoms, with anxiety evidence weaker and sometimes U-shaped — entirely observational, with unvalidated self-reported exposure throughout.

Loneliness raises coronary heart disease and stroke risk ~30%

A meta-analysis of 23 longitudinal cohorts (181,006 participants) found 29% higher CHD risk and 32% higher stroke risk with poor social relationships — funnel asymmetry and random-vs-fixed estimate gaps suggest the true effect may be closer to ~18–19%.

Postpartum depression: screening, SSRIs, and therapy in NEJM practice review

Postpartum depression affects roughly 6.5–12.9% of mothers and is most strongly predicted by antenatal depression — but the headline prevention claim rests on a single small observational study (n=78) with implausibly large effect rates demanding replication.

Dual GLP-1/GIP agonists vs GLP-1 agonists: efficacy real, review weak

Tirzepatide outperforms GLP-1 receptor agonists on HbA1c (up to 2.4%) and weight (up to 22%) in cited trials — but the review itself has no PRISMA flow, no risk-of-bias assessment, and conflicting search descriptions; a separate network meta-analysis corroborates the direction.

EBV mimicry at a specific epitope in a subset of multiple sclerosis patients

Intrathecal antibodies in 6/40 MS patients converge on an EBV BRRF2 motif that cross-reacts with human CNS proteins (RTN2, TRIM71), with 99.36% specificity in a validation cohort of 909 MS cases — applicable to only ~8% of patients, and causality (initiation vs amplification) remains unresolved.

Centenarians' Spanish-flu antibodies show no COVID-19 cross-protection

33 Sicilian centenarians carried ~50-fold higher neutralizing titers against 1918 H1N1 than controls, with zero SARS-CoV-2 cross-reactivity — ruling out antibody cross-protection as any explanation for COVID resilience; N=33 with only 8 infections limits all inference.

Autism incidence rise: mostly broadened diagnosis, not new epidemic

Rutter's 2005 review concludes true autism prevalence is likely 30–60 per 10,000 versus the original 4 per 10,000 estimate, with most of the rise from improved ascertainment and broadened criteria, and no epidemiological support for MMR or thimerosal causation — the broadening-vs-true-rise decomposition remains partially unresolved.

Sodium benzoate, fat accumulation, and aging in worms — hypothesis only

An unreviewed preprint reports fat accumulation and shortened lifespan in C. elegans via SKN-1/Nrf2 signaling — non-monotonic dose effects, missing datasets, and no direct molecular mechanism make the claim hypothesis-generating for a GRAS preservative at best.

Globin digest peptides promote muscle growth across species

Orally administered globin digest increased hypertrophy and performance in zebrafish, mice, and C2C12 cells, with stage-specific peptide activity claimed — no isonitrogenous protein control, small samples, and corporate conflicts of interest; the proposed PHT1/mTORC1 relay is untested.

Exosomal markers of HIIT success: extrapolation, not evidence

HIIT demonstrably raises circulating PGC-1α (63% vs 33% vs no HIIT), lowers IL-6, and increases BDNF/irisin — but no supplied study isolated extracellular vesicles, so any "exosomal marker" claim is indirect until EV-isolation trials are performed.

Sixteen biological-aging markers compared head-to-head in BASE-II

In 1,083 adults with 7.4-year follow-up, Allostatic Load and DunedinPACE most consistently predicted aging outcomes, while first-generation clocks and BrainAge showed no Bonferroni-corrected associations — markers predicted outcomes better prospectively than concurrently, suggesting a measurable pre-clinical window.

Rare FANCI variant in elite athletes: likely a multiple-testing artifact

A stop-gain FANCI variant appeared in 3/101 elite athletes and 0/1,899 controls — but Bonferroni-adjusted p = 0.069, no physiological data on carriers exist, and the authors themselves concede chance cannot be excluded.

Earth-analogue planet survival in binary star systems: no quantified grid exists

No published N-body grid reports habitable-zone planet survival fractions as a function of binary separation and eccentricity — the periastron-collapse hypothesis (eccentricity at large separation can be deadlier than near separation) is testable with open-source tools but currently unquantified.