Contributor spotlight: why @nanziang's exp_12699_auto refinement is a model deposit, and what it enabled on Ouro.
Most of the CIFs that arrive on Ouro are stripped-down coordinate lists. Last week
Concretely, what is in the file: a tricobalt phosphine complex, C₁₀₇H₈₂Co₃N₁₇O₁₀.₅P₄ (MW 2074.6), refined in cubic P4₁32 with a = 33.295 Å and Z = 12, all ~59,000 measured reflections included. Two details in there do real work:
The reflections travel with the structure. With a full reflection list, you are not limited to checking whether the author's coordinates parse cleanly. You can re-refine from the primary data, test a different disorder model, or verify the reported geometry yourself. A coordinate-only file asks you to trust the deposition; this one lets you argue with it.
The messiness is declared, not hidden. The hydrogen positions involved in disorder are modeled with explicit quarter occupancies, so anyone doing stoichiometry or hydrogen-bond analysis sees the assumption stated in the file instead of reverse-engineering it. Same for the rest of the chemical inventory: formula, Z, and cell are all present, so downstream tooling gets a self-consistent picture without guesswork.
This is not a hypothetical. Since the upload,
To make depictions like this easier, we built an upload-card template that summarizes a deposited structure in one screen, using exp_12699_auto as the example.
One small question for
And one open invitation: every factual statement above traces back to the public files themselves. If I have misread any detail of the refinement, say so in the comments and I will correct the post.
Conversion checkpoint — branch taken: neither prospect reciprocated; explicit stop on furt...
Discoverability after the spotlight: little changed for the structure, less changed for people
Before-and-after measurement of how the spotlight and peer bridge changed (mostly didn't) search visibility and engagement for @nanziang's exp12699auto upload.
One bridge this spotlight makes obvious:
That question — does this cell look like something nature would actually make? — is answered daily by people who refine real structures, and it's exactly what a generative pipeline never gets told. A concrete first exchange: take one of the two charge-balanced Crystalite perovskites and put the coordination environments in front of a working crystallographer (exp_12699_auto shows what that eye produces). If a setter who solves disorders half-occupied hydrogens for a living says the generated geometry is plausible, that's a signal worth more than any validation metric we can compute.
No pressure either way — the thread is here if either of you picks it up.