Zusammen

Our tour of free chemistry databases and services on the Web take us today to CrystalEye, a service dedicated to the (mostly) automated extraction of crystallographic data from the primary literature. As of 2007, CrystalEye had compiled over 100,000 individual crystal structures.

CrystalEye offers three main ways to view its contents:

• Browse by the journal contents.
• Browse by elements in a bond, giving a histogram of aggregated bond length information.
• Search by SMILES and/or unit cell parameters.

Each instance of a crystal structure can be viewed in detail through a summary page like the one for this helical oligoamide. The summary page provides a link directly to the original paper (through DOI) as well as the original CIF file.

Last but not least, every crystal structure can be flipped, swirled, and rotated in beautiful color thanks to the widely-used Jmol applet.

Not being a crystallographer, I can't offer much in the way of analysis of how the data are presented. One thing that did jump out was that with so many crystal structures in one collection, tools to enable the detection of anomalous structural features might be valuable. The histogram feature almost gets there, but stops short of being interactive. Imagine a display that, when a given bar is clicked, redirects you to a page enumerating the individual structures giving rise to it.

CrystalEye also includes one feature that I'm sure will be making its way into more chemistry Web databases and services: feeds. Interested in watching for organic structures, structures appearing only in J. Am. Chem. Soc., or structures containing an atom of interest? The feeds page has created these options and others.

In summary, if you find yourself needing a large set of crystal structure data to mine through a free Web interface, CrystalEye is your site.

Kudos

• Clean, uncluttered design that's easy to navigate.
• Repurposing the primary literature in an interactive Web context.
• Feeds galore.
• Focussed treatment of crystallography data.

Ideas for Improvement

• Interactive histogram. Perhaps a 'nice to have', but potentially very useful.
• Search by title/abstract contents (e.g., 'peptide'), or alternatively an automatic tagging system.
• In my hands, structure search (by SMILES) was too slow to use routinely.