POLYMER GENOMICSNewsroom·claim dispatches
§ NEWSROOM · 12 claim dispatches · all evidence on the public record
Newsroom
Every in-silico experiment posts a formal claim — premise, operation, statistic, inference rule, conclusion — with the outcome flagged. Negative results included by design. Click a card to open its full claim record.
8Positive2Qualified2Negative
Filter
MAY 18 · EXP 22
CpG erosion chronometer recovers transposon ageA per-TE methylation-based chronometer dates L1HS and AluY copies to within 1.4 MYA of phylogenetic estimates, validating mC erosion as an evolutionary clock.
r = 0.81 · MAE 1.4 MYAPOSITIVEMAY 12 · EXP 21GC-arrangement decomposition · LP proofRoll, twist, rise, and groove geometry are 60–87 % arrangement-determined, isolating GC-independent biophysical signal genome-wide.
Arrangement variance up to 87 %STRONG +MAY 04 · EXP 19Replication-timing partial correlationsZ-form propensity emerges as the top biophysical correlate of replication timing once GC and TE density are controlled for.
Partial r = 0.100 · rank #1POSITIVEAPR 28 · EXP 18Gene-flanking deformability predicts expression breadthDeformability of 10 kb flanking regions partially correlates with cross-tissue expression breadth independent of GC and CpG content.
Partial r = −0.345POSITIVEAPR 20 · EXP 17Crossover hotspots predicted from arrangement biophysicsA 7-feature arrangement-only model (no GC) discriminates meiotic crossover hotspots from background with AUROC 0.827.
AUROC 0.827 · GC-controlledSTRONG +APR 12 · EXP 16HLA expression mismatch from non-coding biophysicsNon-coding biophysics divergence between HLA alleles predicts expression-level mismatch better than coding identity alone.
r² = 0.41 · n = 188 pairsQUALIFIEDAPR 06 · EXP 23Meiotic groove-width effect size · cancer nullGroove-width is strongly arrangement-dominated in meiotic hotspots (Cohen d = −0.41) but does not transfer to cancer breakpoints in this cohort.
Meiotic d = −0.41 · cancer nullQUALIFIEDMAR 28 · EXP 15Stacking energy distinguishes essential genesMean stacking ΔG₃₇ across gene body separates essential (LOF-intolerant) from non-essential genes after constraint correction.
d = 0.62POSITIVEMAR 20 · EXP 11Biophysical similarity does not predict shared regulationA 64-feature biophysics fingerprint fails to recover known co-regulated gene clusters above chance.
AUROC 0.51 (n.s.)NEGATIVEMAR 12 · EXP 09Curvature does not encode promoter strength aloneIntrinsic curvature in core promoters does not predict TPM rank above the GC-content baseline.
ΔAUROC = +0.01 (n.s.)NEGATIVEFEB 28 · EXP 03TE silencing predicted from CpG context + ageA two-feature model (CpG density × TE age) discriminates silenced from active TEs in lymphoblastoid lines at AUROC 0.84.
AUROC 0.84POSITIVEFEB 04 · EXP 01Stacking ΔG₃₇ is calibrated to nearest-neighbor parametersWithin-database calibration of stacking ΔG₃₇ against SantaLucia 2004 nearest-neighbor parameters yields r = 0.998 on a 5,000-region benchmark.
r = 0.998STRONG +