The only precondition is equilibrium. That's definitely not true for a cell, perhaps not even true for in vitro, but close enough to make the point that each component adds the possibility of another phase.

f+p = c+2

For in-vitro c is small, typically 2 or 3. For in-vivo c can be huge, upwards of 10,000 for a typical mammalian tissue: So the number of phases in-vivo rises.

The Gibbs rule can be viewed as the thermodynamic justification for the statement that one needs pure substances to get crystals

The short answer is no, because of Gibb’s rule.

This remark is especially on-point for structural biologists lately. Augmentation is always enshittification to some degree, especially when extrapolated out to several orders of magnitude beyond simulation limits of a model full of dubious assumptions to begin with. E.g: bsky.app/profile/fran...

Water networks in xtal structures is something I have been analyzing extensively, but until I saw your post I hadn’t looked at cryoEM. May I ask what temperature was used? Network size goes up linearly with decreasing T. My DMs are open.