Like DNA, RNA can convey information, but unlike DNA which is always the same boring helix, RNA can twist up into complicated 3D shapes as proteins can; and that's important because like proteins, shape determines what they can do, and one of the things RNA can do is act as a biological catalyst. And the human genome transcribes 30 times as much RNA as it does for proteins. Unfortunately experimental methods to find the 3-D shape of RNA molecules have proven to be even more difficult than it was with proteins, so only a few have been found experimentally, less than 1% of the number found for proteins. Years ago computer programs were developed to predict the 3D RNA structure from its 1D nucleotide sequence, but the resulting prediction of the 3-D position of atoms was in error by 16-Å (a sulfur atom is about one Ångström wide), that's too inaccurate to be useful for drug discovery and is far worse than the most recent protein structure program that predicts the proteins 3-D shape from its 1D amino acid sequence with errors of only 2 Å. We've been stuck at that 16-Å figure for a very long time but in yesterday's issue of the journal Science researchers report on a new neural net program called "ARES" that reduces that error from 16-Å to 12-Å, still too inaccurate to be very useful but it's significant progress.
To train ARES they used 18 of the few RNA molecules in which the 3D structure had been determined experimentally between 1994 and 2006 and a far greater number of incorrect structures so that the program could learn what worked and what didn't; they then tested how well it was doing by asking it to predict the structure of the few other RNA structures found experimentally between 2010 and 2017. And they found it was doing pretty well, although more improvement is needed. AI neural nets improves 3D RNA structure prediction <https://science.sciencemag.org/content/373/6558/1047> John K Clark See what's on my new list at Extropolis <https://groups.google.com/g/extropolis> c88 -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to everything-list+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CAJPayv08k6cs%3DypEOQCshaVCskOTsy85WHUTz_gBhvYpKDd%3DHQ%40mail.gmail.com.