19 comments

[ 3.0 ms ] story [ 53.3 ms ] thread
I like this project a lot (and use it ocassionally in my PhD). I've tried this full profile Rietveld-esque tool [0] before, and it works pretty well for being so simple (at least from a UX point of view).

What I would like to see (and would program myself if I was a better programmer and knew more of the math and physics behind XRD) is a libre search-and-match program, using a peak database (in reality, multiple databases for different X-ray sources) constructed from the CIF files in COD.

[0] http://cod.iutcaen.unicaen.fr/

I think that's where commercial programs make their money--in phase identification. JADE comes to mind
What are some other good scientific databases compiled from the open literature?

Are there any best practices for the construction of similar databases?

For crystallography specifically, there's ourselves (Materials Project), OQMD, AFLOW, Materials Cloud, JARVIS, and a number of more specific (but no less important) specialized databases. There are also a number of commercial offerings.

Best practices are incredibly difficult. We're trying to establish a common API currently (https://github.com/Materials-Consortia/optimade) that can be adopted by all database providers. How the data is stored behind the scenes is something that ends up being very specific to how the data is generated and what its applications are. We're definitely better as a community than we were ten years ago, but there's a lot of work to be done here.

In terms of scientific databases outside crystallography/materials science, Nature's Scientific Data is a good open-access journal to peruse: https://www.nature.com/sdata/

There's the classic, Protein Data Bank. A number of people in the field saw it as a moral imperative. It's interesting how open different disciplines are to openness
Love COD, very nice to see them on the front page :-)

Along similar lines, we maintain an open database of crystal structures ourselves at the Materials Project (materialsproject.org) at Lawrence Berkeley National Lab, which includes high-quality calculations of the electronic structure of each crystal structure so that you can do queries for the specific properties you need for your application. We're currently working alongside the Crystallography Open Database and several other databases to develop an API spec so that all similar databases can be queried in a common way: https://github.com/Materials-Consortia/optimade

If it's ok to offer an ad here, we're actually looking for someone to join our team. The calculations we offer at the Materials Project are the results of millions of CPU hours of computation, so it becomes very important that we're able to synthesize this data in a way that's approachable and understandable and then share this data effectively with other researchers, students and members of the public who might want to use it. We need help with this and we're hiring a web developer at the moment: https://news.ycombinator.com/item?id=23381358

I must admit I didn't expect OPTIMADE to hit HN before it hit arXiv!
The materials project has experimental and theoretical structures. On the experimental side, how does it compare to ICSD? How up to date is it in comparison? I love the API for the materials project!
> I love the API for the materials project!

Ah, so happy it's useful to you!

We're working on a new API internally too (based on FastAPI) that will hopefully bring better documentation along with it, so stay tuned for improvements.

> On the experimental side, how does it compare to ICSD?

We have pretty good coverage of ICSD and other experimental databases, and we continue to process and calculate new materials as they're discovered. We also calculate ordered approximations of disordered structures too, but this is an area where we could improve.

We also provide a capability where users can upload crystal structures we don't have and we calculate those too (with credit going to the original uploader).

> Open-access collection of crystal structures of organic, inorganic, metal-organics compounds and minerals, excluding biopolymers.

"biopolymers" links to the RCSB PDB (http://www.rcsb.org/), which stores biological macromolecular structures.

On the hunt for file formats to pass I ended up implementing a parser for the CIF format https://github.com/EliotJones/BioCif

It's great that these projects to create open data stores exist.

As a biologist, a field where open databases are the norm and deposition is required for publication, I was very surprised when I found out that the standard database of small molecule crystal structures was NOT open.
What we also really need is an open antibody database, with links to buying antibodies and info like affinity, gene sequence and epitope binding. I know measuring affinity and epitope mapping can be hard but this is really necessary for for example COVID-19 diagnostics and vaccines.