a public research recordlast updated 11 September 2026
The science

Six numbers, and what each one is allowed to claim.

Every figure below is measured or derived from a named source. Where two of them disagree, both are shown.

The structural argument

The variant sits in the channel's N-terminal domain, the cytoplasmic foot that hangs inside the cell before the membrane-spanning machinery begins. In the wild-type channel, Arginine 104 forms an ionic salt bridge with Aspartate 84. Substituting glutamine removes that positive charge, replacing the rigid ionic link with a neutral polar hydrogen bond. That contact was identified in the deposited cryo-electron microscopy coordinates and confirmed across 1.41 million atoms in all-atom molecular dynamics.

Matched All-Atom Simulation · 1,410,000 Atoms OpenMM Double Precision

Simulating the complete channel in a physiological POPC lipid bilayer with 120 mM NaCl reveals the physical consequence of the mutation:

DEKA Pore Radius
5.30 Å
Wild type measures 5.34 Å. The selectivity filter does not collapse.
Pore Hydration
12.8 waters
Wild type holds 12.7 waters. The conduction pathway remains fully wetted.
Contact Excursion
3.58 Å → 7.33 Å
Mean 3.58 Å, maximum excursion 7.33 Å. The cytoplasmic foot uncouples into a loose hinge.
01 · one value, measured in a dish
68.3
percent
± 6.1 · n = 34 cells
0one working copy, on its own = 100

This is the only functional measurement of this variant that exists. Everything downstream of it on this site is inference, and it is labelled as such. PMID 35305865

02 · two rival explanations

Does the broken copy simply fail to arrive, or does it interfere with the good one?

If the broken copy simply never arrives45.8%
Measured, rescaled to a two copy heart31.3%
14.5 points, unexplained

That gap is the honest centre of the whole project. The Turnover Observability theorem formalises that if whole-cell current falls below the proportion of surface wild-type protein (I/I0 < Sw/S0), channel conductance per surface unit is mathematically forced to drop. Because the all-atom simulation confirms that the channel core remains folded and fully hydrated, the mutant channel is expected to pass endoplasmic reticulum quality control and reach the surface, pointing to coupled gating or dominant-negative suppression rather than simple trafficking failure.

What would settle it

An assay that reads each copy separately rather than the total. It is costed in full on the experiments page, using Lys-C mass spectrometry and matched patch clamp, with its internal control stated in advance.

Divisor 2.184, not 2: two working copies were measured at 218.4 percent of one, so 68.3 / 2.184 = 31.3 and 100 / 2.184 = 45.8. Corrected 6 August 2026. The retired baseline and the retired comparator travel together, so quoting 31.3 against 50 is a third wrong answer rather than a partial fix.
03 · a filter, and what survives it
85.1
percent
of apparent off target risk removed
counted by sequence similarity alonecould actually be edited

Asking whether an off target site could actually be edited, rather than whether it merely looks similar, removes most of the apparent risk. What survives is not zero: 22 missense and 26 protein-changing candidate off-target sites remain in the genome, with zero sites within two mismatches.

04 · a substrate, against the tissue where the therapy works
Brain cortex, where this approach already works in children1.388%
Human heart, across 827 samples0.0045%
not visible at this scale, which is the finding

About 300 times smaller than the tissue where the approach works. Redirecting every discarded transcript buys 1.06 fold where the route needed about 1.5, so it is closed by a substrate that is not there rather than by a delivery problem someone might later solve.

05 · a count, against my own method
1 / 4
variants already known to break this channel were caught by the predictor I had been relying on

A high score still means something. A reassuring score means nothing at all in this part of the gene, which closed the route this project had spent months on. Four is a small denominator, and the paper says so throughout.

06 · one distance, measured two ways, against a cutoff
Atom to atom, the real contact3.79 Å
Centre to centre, what the model measures9.22 Å
the cutoff

The contact at this position is real and close. Measured centre to centre it falls outside the cutoff, so a centre distance model cannot see it at all. About 9 percent of contacts in this domain are lost the same way, and the long side chain ones go first.

Where this project's own numbers disagree

How buried position 104 is inside the assembled channel has five different values across five analyses, depending on the method and on whether the domain is measured alone or in the whole protein. Where measurements disagree, no single value is quoted anywhere on this site. All of them are listed.