>November 24, 2004 (USA): These work units are not "bad" work units at all, >but are what we call work units that don't converge. >Rosetta works by attempting to fold a protein sequence many, many times. >It does this over and over again, each time trying different ways of folding it. > Sometimes, though, it just can't find a way to fold a protein in a given amount of time. >When this situation occurs, we give up on the work unit, > send what you have completed so far back to the server and the agent will download the next work unit. >We found a bug where this information wasn't getting sent back to the server under this scenario. >Thus points weren't being awarded for the work done.
>To summarize, work units that fail to converge are normal, > so there will be times where you may see a work unit cranking away for a couple of hours with no progress. >It also means that the work unit will end prior to reaching 100% > because we detected this situation and end the work unit early. >This information is valuable and helps in determining the proper fold for a particular sequence.
>We have identified the problem and are currently testing the fix. >We plan to deploy the code on Friday, November 26, 2004 (USA) and will post an update upon deployment.