Research – Biology and chemistry

A map of everything inside a human cell

A catalog of every kind of thing in a human cell, each one written as a JSON body of physics and chemistry. It is the data a cell simulator will be built from.

StartedNov 10, 2026
Last updateNov 10, 2026
kinds of thing catalogued, from water to viruses195

Question

What would a simulator need to know about every part of a human cell? Size, mass, how many copies, how it moves, what it is made of, what it does. The goal is one consistent record per part, in numbers with units, before any simulation code is written.

Approach

Every entity gets a line in a catalog and a body file. All bodies share one base template: identity, location, physics, chemistry, state and tumor notes. Extra blocks are added only when an entity has the matching trait. A small Python tool stamps blank bodies, validates them against the template and counts progress.

Piece What it holds
Base block parent compartment, shape, size, mass, copies per cell, net charge, mobility (fixed, 3D or 2D diffusion, motor-driven), composition, processes, state variables, tumor notes
Traits (13) membrane, lumen, genome, polymer, catalytic, transporter, signaling, encoded, soluble, lifecycle, condensate, cell type specific, pathological
Kinds (6) compartment, structure, complex, macromolecule, small molecule, feature (a region or mark on something else)
Reference cell a generic dividing human cell in culture, roughly HeLa-sized: 18 µm across, about 3 ng. Every per-cell count assumes it

Every number is a quantity: a value, a unit from a fixed list, and a basis. The basis says how much to trust it: textbook (widely quoted), estimate (order of magnitude or worked out by arithmetic) or varies (depends strongly on cell type; one typical case). Ranges and notes are optional.

A body can also have state variables a simulator would track. The sodium-potassium pump, for example, carries its conformation and a cycle rate of about 50 per second, plus a transport block (3 sodium out, 2 potassium in per ATP), a catalysis block with kcat and Km, and its gene.

Results

A stub is a blank body. A draft is filled from recall but not yet checked against a source. Nothing is marked reviewed yet.

Group Entities Draft Stub
Cell (whole cell, cytosol, outside space) 3 2 1
Surface (membrane and what sits in it) 22 2 20
Organelle 39 1 38
Structure (cytoskeleton, machines, condensates) 22 3 19
Genetic (DNA, RNA, packaging, marks) 22 1 21
Protein 28 1 27
Small molecule 47 2 45
Abnormal (invaders, damage, tumor oddities) 12 1 11
Total 195 13 182

The 13 drafts so far: cell, cytosol, plasma membrane, Na+/K+-ATPase, mitochondrion, ribosome, microtubule, stress granule, mitochondrial DNA, p53, ATP, calcium ion and extrachromosomal DNA (ecDNA, the loose DNA circles found in many tumors).

By kind, the catalog has 51 macromolecules, 47 small molecules, 38 structures, 35 compartments, 16 complexes and 8 features.

Findings

  • One shared template with optional trait blocks keeps 195 very different things comparable. A calcium ion and a mitochondrion answer the same base questions.
  • Tagging each number with a basis matters. Many cell figures vary tenfold or more between cell types, and a simulator needs to know which values are soft.
  • Fixing one reference cell up front avoids mixing counts from cells of very different sizes.
  • Tumor notes sit on every body, so cancer-relevant changes are recorded next to the normal physics rather than in a separate file.
  • Species names that match a catalog id act as links, so composition lists double as a graph of what is made of what.

Next

  • Fill the 182 stubs.
  • Check drafts against sources and move them to reviewed.
  • Use the bodies as the input data for a cell simulator.