Science supplement index · Public Blogger finding

A unit correction changes a probe-length calculation, but does not prove long probes fail

Reviewing the supplied test of Hybridisation Specificity Is a Ratio uncovered a unit inconsistency. Correcting it preserves a shrinking temperature gap in the chosen model, while changing the numerical values substantially. The result is a repaired physical-chemistry illustration, not experimental proof that a long DNA probe cannot distinguish a single mismatch.

What was corrected

The supplied program assigns its enthalpy terms in kilocalories per mole and entropy terms in kilocalories per mole per kelvin. It sets the gas constant R to 0.0019872 in matching kilocalorie units. The melting-temperature function then multiplies enthalpy and entropy by 1,000, but leaves R unchanged. That makes the concentration contribution 1,000 times too small relative to the entropy term.

The repair changes one line: use T_m = ΔH / [ΔS + R ln(C_T/4)] − 273.15 with all three energy quantities in kilocalorie units. Equivalently, convert all three to calories. The concentration convention assumes two equally concentrated, non-self-complementary strands. SantaLucia’s primary thermodynamics paper explains the concentration factor and consistent gas-constant units; it does not validate this supplied parameter set.

Every other supplied choice stays fixed: total strand concentration of 10⁻⁹ M, the mean bulk enthalpy and entropy terms, the initiation terms, the fixed mismatch penalties, and all 55 integer lengths from 6 through 60. This is a correction to an existing calculation, rather than a newly designed experiment.

What the same model now gives

The gap means the perfect-duplex melting temperature minus the mismatched-duplex temperature. It remains positive and strictly decreases across the supplied grid. At length 6 the corrected model predicts both temperatures below 0°C, so even the short-probe endpoint must not be read as an ordinary assay recommendation.

The original NumPy calculation, with only the unit line changed, agrees with the standalone public reproduction at all 55 lengths. An independent check also converts enthalpy, entropy and R together from kilocalories to calories: both temperature predictions stay unchanged to approximately 1.2 × 10⁻¹³ °C. These are numerical consistency checks, not confidence intervals or laboratory uncertainty estimates.

What is still unproved

The supplied program multiplies its mean terms by length n. It does not enumerate a real sequence’s n−1 neighboring stacks. Its fixed mismatch penalty is assumed rather than fitted to a particular mismatch and sequence context. Salt effects, secondary structures, probe surfaces and an assay’s detection process are absent. We retain that bulk model transparently instead of presenting it as a calibrated sequence-specific nearest-neighbor prediction.

A smaller positive melting-temperature separation is not a demonstration of zero discrimination. No detection threshold, measurement noise, binding selectivity or washing protocol is modeled here. Establishing practical failure would require appropriate sequence-specific predictions and measurements under defined assay conditions.

The defensible finding is limited: correcting the units changes the temperatures, and the chosen bulk model still produces a decreasing mismatch temperature gap with length. The original claim of inability remains unsupported. Find the original proposal and related evidence in the Incoprea research notebook.