How to read a locus

A locus is read in one deterministic pass. From its sequence the engine computes γ (stacking stiffness), which fixes the R19 threshold scale; places the element in its A4 coordinate (shell, nearest anchor and strength, loops, anchor-relative helical phase); decides whether the R19 well is bistable (can-fire); and returns five channels. Walked here on human_SOX2, γ = 1.287315.

This chapter is the method, end to end, on one worked locus. Reading is mechanical and order-free: γ from nearest-neighbour stacking sets the threshold scale (spinodal ∝ γ^1.5, barrier γ²/4); the A4 coordinate places the element relative to its nearest anchor and reads the helical phase between them; the R19 switch reports can-fire while on/off stays runtime; and the five channels (material, set, state, dwell, environment) are the output of that single read, not five separate analyses.

Step 1 — What is γ, and what does it fix?

γ is the stacking stiffness of the duplex: γ = −mean nearest-neighbour stacking ΔG over the sequence, using the SantaLucia 1998 table. It is read mechanically — no fitting, no free parameter — and re-derives identically every run. For human_SOX2 (302,512 bp) the engine reads γ = 1.287315 at GC = 0.396.

γ fixes the scale of the R19 switch threshold, not the trait. From γ alone the switch's spinodal (the tilt that collapses one basin) is |h_sp| = (2/3√3)·γ^1.5 and its barrier is γ²/4. For human_SOX2 that is spinodal = 0.5622 and barrier = 0.4143. Because γ is an affine read of GC (corr(γ,GC)=0.998 across the census), γ carries composition, not lineage.

Step 2 — Where does the element sit? The A4 coordinate

The same read places the element in a four-part coordinate, so a locus is read structurally rather than as a flat string. The four components are: the shell (soft / mid / stiff composition band) the element falls in; the nearest anchor and its strength; the loops co-anchored with it, built from real motors parsed from the region's feature table; and the anchor-relative helical phase.

The phase is the physically meaningful part. Using the locked B-DNA geometry — rise 3.4 Å and twist 34.29° per base pair — the engine asks whether the element and its nearest anchor fall on the same rotational face of the helix (and could contact) or on opposite faces (and cannot). On the worked lactase promoter the read is γ 1.315 · mid-shell · anchor 1065 bp · 1 loop / 5 motors · face 0.44 → contact False. The same grammar reads human_SOX2; where a window is gene-poor the loop read is flagged open and names the dataset that would close it.

The helical face is read as geometry, not function. Across twelve cross-kingdom regions a real TSS sits on the same helical face as its nearest anchor at a rate of 0.337, against an analytic chance of 0.340 (246 motor–anchor pairs, permutation z = −0.09) — indistinguishable from chance. So contact_competent is a deterministic Layer-1 geometry read, never a functional-contact claim; realized looping depends on runtime factors (bound factors, cohesin) that are Layer-2. geometry / chance-level stress test → repro

Step 3 — Can it fire? R19, and why on/off is not readable

The R19 switch is a double-well in a state variable s, with dynamics ṡ = −(s³ − γs − h). For γ > 0 the well is bistable — it has an ON basin and an OFF basin — so the switch can fire. human_SOX2, at γ = 1.287315, is bistable. Which basin it rests in (on or off) is set by the environmental tilt h, not by γ, so the on/off STATE is runtime and is not read from the sequence. This is the boundary in miniature: γ reads that a switch can fire; it is blind to whether it is firing.

Step 4 — What does one read return? The five channels

One read returns five channels at once, not five analyses. MATERIAL is γ and its threshold scale. SET is the presence/copy-count of switches a genome carries. STATE is on/off (runtime-flagged). DWELL is how long a switch stays on (∝ γ^1.5 × dosage; direction admissible, absolute open). ENVIRONMENT is the tilt h the environment writes via methylation (the CpG handles) and temperature, within a fixed SET. Parts III–V take each channel in turn; here they are simply the output of the single read above.

In one line: read γ → threshold scale, place it in the A4 coordinate, ask R19 can-fire, and report the five channels — every value mechanical and reproducible, the on/off sign and amount left explicitly to runtime.