A protein usually arrives on screen as a bundle of ribbons: loops, spirals and arrows rendered in colors no cell has ever requested. The source can look considerably less glamorous.
Here are the first three atoms in a model of crambin, a small protein found in seeds:
ATOM 1 N THR A 1 17.047 14.099 3.625 1.00 13.79 N
ATOM 2 CA THR A 1 16.967 12.784 4.338 1.00 10.80 C
ATOM 3 C THR A 1 15.685 12.755 5.133 1.00 9.19 C
That is not a program. It is a map.
The first row identifies a nitrogen atom in threonine, the first amino-acid residue in chain A. The three decimals in the middle are its x, y and z coordinates in ångströms. The next row places a carbon atom nearby. Continue for 327 ATOM records and software can reconstruct the full model.
I downloaded entry 1CRN from the Protein Data Bank and parsed those fixed-width columns. The file contains 46 residues and 327 modeled protein atoms. Its coordinates span roughly 27 by 21 by 27 ångströms, so the whole thing would fit inside a cube about 2.7 nanometers wide. The protein is small enough to be approachable, although “approachable” is doing some work when the approach begins with crystallography.
The picture is hiding in the numbers
A molecular viewer reads the same facts a patient person could extract with a ruler and a column guide: the atom and residue names, chain, position, occupancy, and a temperature factor related to how sharply or diffusely the atom appears in the experimental map.
From those positions, the viewer draws bonds, groups atoms into amino acids, traces the protein backbone and replaces the resulting thicket with a ribbon. None of those prettier representations is literally stored in the three lines above. The spiral appears because the coordinates place successive atoms in the geometry of a helix.
This is a useful correction to the idea that scientific data must look like the thing it describes. A weather file does not contain a cloud. A music file does not contain a trumpet. A protein coordinate file contains enough measured structure for software to build a view.
It is also not a photograph of a protein. The PDB describes its primary records as atomic coordinates plus information about the structure, sequence and experiment. The coordinates are a model derived from experimental observations. Some structures have missing regions, uncertain positions or several plausible conformations. The tidy ribbon is an interpretation with excellent posture.
Seven structures became 261,991
The format belongs to a larger, unusually durable act of scientific sharing. The Protein Data Bank was announced on October 20, 1971, and its first archive held seven structures at Brookhaven National Laboratory. This month marks its 55th anniversary.
When I queried the archive today, it returned 261,991 current entries. They include proteins, nucleic acids and large molecular assemblies. Researchers deposit structures so other people can inspect them, compare them and build on them rather than treating every set of coordinates as private laboratory furniture.
Crambin entered that archive in 1981. Its legacy .pdb file is charmingly rigid: every field lives in assigned character columns, an arrangement that makes a line readable once you know where to look and fragile once biology becomes too large for the boxes. The format has had a remarkable run, but it is now the old doorway.
The archive's modern PDBx/mmCIF format names fields explicitly and can represent more atoms, residues and chains without the legacy limits. In July 2027, new entries will stop receiving four-character IDs and legacy PDB files. Crambin's familiar 1CRN will remain, but future structures will arrive through a wider door.
That transition is a small history of computing in miniature. First, squeeze a protein into columns designed for the machines you have. Then keep the shared convention alive long enough for science to outgrow it.
The ribbon on screen is beautiful, but the plain-text file may be the more consequential object. It can be downloaded, checked, copied and reinterpreted decades later. Three hundred twenty-seven lines are enough to turn a seed protein into a public coordinate system, and the rest is drawing.