Those of us who were trained in physics or chemistry tend to think of a cell as a bag, mostly water, with some active molecules floating around randomly in it. This is really misleading.
Chemical engineers, for example, often envision a "well stirred reactor." (Physicists do too, the difference being that they often do it without realizing it.) In this model, every molecule is equally likely to be anywhere in the reactor, which we take to be a cell. Maybe we have one pot for the inside of the nucleus and one for the cytoplasm, but we imagine them individually to be well stirred. This makes mathematical modeling really straightforward. For example, the rate at which two types of molecule react is proportional to the product of their concentrations in the cell, times the likelihood that they will react if they bump into each other. From here it's easy to write a set of simple differential equations describing how the number of each type of molecule changes over time. Easy, but wrong--or at least oversimplified.
Consider messenger RNA, which is an inverted copy of the genetic information in DNA. In eukaryotes like us, after the RNA is transcribed in the nucleus, it moves to a ribosome out in the cytoplasm of the cell, where it is translated into protein. I used to imagine this just like a well stirred reactor: the freshly minted RNA diffuses randomly around the nucleus, some of it leaks out through holes in the nuclear membrane into the cytoplasm, and there it randomly bumps in to a ribosome to begin translation.
A big problem with this image is that in reality the RNA is virtually never alone. Everywhere it goes, it is accompanied by an entourage of proteins that guide it to its next press conference, keep away the paparazzi, and so forth. Nobody gets in without an appointment.
Even as the RNA is being transcribed, proteins modify it with tags that identify it as a protein-coding RNA, rather than, for example, an RNA involved in regulating gene expression. It is quickly shepherded to the splicing apparatus, where more proteins remove some sections and splice the others together to form mature messenger RNA. Special bodyguard proteins then accompany it through the security gate at the nuclear pore. Although its name suggests a simple hole in the nuclear membrane, the nuclear pore doesn't let any molecules in or out without one of these bodyguards. The RNA may then be escorted directly to a ribosome to begin translation. Alternatively, it may be shunted to a holding area within the cell to await further instructions. If its sequence matches one of the regulatory RNAs, which has its own protein entourage, the enforcer from the regulatory entourage executes the messenger RNA, cutting it to pieces before any translation. So it goes.
Proteins also spend much of their time in direct contact with other proteins, beginning even during their translation from RNA. These transient associations aren't easy to measure, but biologists are continually improving their tools for marking individual proteins, imaging their locations, and recording who they associate with. These tools reveal that the metaphor of the cell as a machine--imagine RNA processing as a production line--is a much more potent than is the metaphor of the well-stirred reactor.