SevenBeat

Some proteins hold on to each other with two helices wound round one another, a coiled coil. Their sequence keeps a meter of seven beats, named a to g. In 1953 Francis Crick worked out why: an α-helix turns about 3.6 residues per turn, so seven residues make almost exactly two turns, and the a and d beats line up as a stripe down one side of the helix.

Toy model A simplified picture to show one idea. Not a measurement, and not a prediction. Of a coiled coil: a helical wheel (each helix seen end-on, seven beats round a circle) scored by rules of thumb, not by a model of real atoms.

Write two strands

Pick a letter, then tap a beat to put it there. Each strand is one heptad (seven beats), repeated 4 times, so 28 residues long.

Strand A

Strand B

The rules of thumb

Oily core. The a and d beats of both strands face each other. Oily side chains there (L, I, V) pack together away from water.
Charged edges. The e and g beats sit at the edge of that seam. A g on one strand faces the e of the next heptad on the other strand. Opposite charges (E with K) pull together; like charges push apart.
The other beats (b, c, f) face the water. Here they do not score.

These are rules of thumb (a toy model), not a prediction. A real chain also depends on how well the helix forms, how many strands come together, and much more. In 1993 Erin O'Shea, Kevin Lumb and Peter Kim used exactly this edge rule to design two peptides that stay unfolded alone but pair once mixed. They called it peptide "Velcro". Try it: make strand A all E at e and g, strand B all K, and compare each strand with itself.

Find the beat in a real coiled coil

GCN4 is a yeast protein that switches on genes. Two copies hold each other by a leucine zipper, one of the first coiled coils solved to atomic detail (O'Shea, Klemm, Kim and Alber, 1991). Start, then tap the residues you think sit on the d beat. Then reveal the register.

Honest notes

The register is found by one simple rule: try all seven ways to label the residues a to g, and keep the one with the most oily side chains (L, I, V, M, F) on a and d. Real register finders weigh much more.
The sequence and the zipper's range come from UniProt's reviewed entry; the confidence comes from the AlphaFold DB model, which is a prediction for one chain on its own. The partner strand is not in the model.
Crick's 3.6 residues per turn is for a straight helix. In a coiled coil the helices bend round each other, so a heptad fits about 3.5 per turn.

Sources

Crick FHC (1953). The packing of α-helices: simple coiled-coils. Acta Cryst 6:689–697. doi:10.1107/S0365110X53001964
O'Shea EK, Klemm JD, Kim PS, Alber T (1991). X-ray structure of the GCN4 leucine zipper, a two-stranded, parallel coiled coil. Science 254:539–544. doi:10.1126/science.1948029
O'Shea EK, Lumb KJ, Kim PS (1993). Peptide 'Velcro': design of a heterodimeric coiled coil. Curr Biol 3:658–667. doi:10.1016/0960-9822(93)90063-T
Gonzalez L, Woolfson DN, Alber T (1996). Buried polar residues and structural specificity in the GCN4 leucine zipper. Nat Struct Biol 3:1011–1018. doi:10.1038/nsb1296-1011
Lupas AN, Gruber M (2005). The structure of α-helical coiled coils. Adv Protein Chem 70:37–78. doi:10.1016/S0065-3233(05)70003-6
UniProt entry for yeast GCN4 and its AlphaFold DB model (both CC BY 4.0): named after Start.