Voting period is over. Please don't add any new votes. Voting period ends on 26 Dec 2015 at 04:42:00 (UTC)
The human DNA clamp protein (PCNA) assembled in a trimer, rendered from PDB: 1AXC.
Reason
I made this image in 2006. We now have many thousands of similar rendered protein structure images, many of which are of equally pretty and symmetrical molecular complexes, and this particular image is of unimpressive technical quality by current standards. (Compare this image of a very similar protein complex from the same article: File:1CZD.png.) I just made an incidental edit to an article it's used in and was reminded of it, and was surprised to see it's still an FP.
Comment Can we rerender and replace? I mean, it's an important enough protein to have its own article, so it's not like any other protein could replace its encyclopedic value. Adam Cuerden(talk)17:33, 14 December 2015 (UTC)[reply]
@Adam Cuerden: Every human protein has its own article :) And the Protein Data Bank - the source of the data for images like this - has over 100k structures in it. I considered just updating it, but even a modern rendering would be very ordinary; it's basically a historical accident that this one became the featured example. I don't really know much about the FP process, but I suggest looking at Evolution and evolvability's work for high-quality protein images. Opabinia regalis (talk) 18:29, 14 December 2015 (UTC)[reply]
I don't know, it's a good example of β-pleated sheets and α-spirals, at a glance, and also of modular proteins. And, honestly, given good renders, I don't think FP would be against every human protein having an FP, if they had decent articles, good images, and were nominated in sets so that we don't have - what is it, ~ 20,000 proteins nominated? Though, I suppose, the exact structure will only be known for a certain number. What's that at now, a couple thousand? Adam Cuerden(talk)20:20, 14 December 2015 (UTC)[reply]
Hmmm, I hadn't thought about it that way (and to be honest, most have crappy articles). There are about 28k structures in the PDB listed as human proteins, but there's a lot of redundancy - looks like around 2400 unique proteins. I know the EBI did an automated batch of protein renders a few years ago and uploaded the results to Commons, but those are looking rather dated now too. I left a note at WT:MCB for their thoughts. Opabinia regalis (talk) 21:14, 14 December 2015 (UTC)[reply]
Well, let's see... since this is a DNA replication protein, it might make sense to start with a complete set of known, structurally-well-defined proteins involved in DNA replication, or just all proteins that act directly on DNA (possibly excluding mitochondrial proteins). I think our largest set is around a hundred or so; don't see why you couldn't reasonably go that level, though that count would need to include any secondary views of the proteins. Adam Cuerden(talk)11:44, 15 December 2015 (UTC)[reply]
Thank you for the kind words Opabinia regalis. I like the idea of doing some general updating of important protein images, Adam Cuerden. I have some PyMOL scripts that I can run to do some cleaner renders of structures. However, really good images will require some manual decision making on what features to highlight (active sites, key residues, substrates, cofactors, oligomers etc) which is sadly harder to automate. If there is interest, I'd happily make some sets of images in the same fundamental style (e.g. Enzyme, Theta_defensin or Plant_lipid_transfer_proteins). I can open source my PyMOL scripts if that's useful to the projects too. T.Shafee(Evo﹠Evo)talk12:25, 17 December 2015 (UTC)[reply]