Try to pull the plug on your computer while processing - you will find that no work is lost.

Unfortunately I doubt work units can be made any smaller as each encompass a single protein. When proteins fold each and every amino acid contributes to its characteristic folding pathway - therefore the whole peptide is required for the calculation. A change to a single amino acid in even the largest of proteins can render it inactive and alter its folding characteristics - the basis for many diseases. This has always been the problem in determining how proteins fold and their shear size requires such high computational requirements.

Some useless info on folding pathways for those interested...

A peptide is a set of linear amino acids - primary structure.
Peptide regions fold into distinct motifs or shapes: alpha helicies, beta sheets, aba barrels etc - secondary structure. This is also determined by disulphide bridges that are formed between cysteine residues within the same pepdite.
The overall shape for a single folded peptide is termed the tertiary structure.

Between seondary and tertiary structure an event termed hydrophobic collapse occurs where all water-hating regions of the peptide are forced to the inward region of the protein and all water-loving amino acids are positioned to the outside of the protein. This event may be spontaneous however may be aided by helper proteins called chaperones - not analysed in this project.

Some folded proteins also contain multiple domains - regions of grouped tertiary structure or may comprise of two similar proteins or subunits that function as one - quarternary structure - also not analysed in this project.

Considering that combinations and permutations that are possible for each peptide, I am suprised how efficient Rosetta is in obtaining a result.

I just hope patience prevails and the team fix the agent soon...

Cant remember exactly where it was but one of the links explains that the expected time is ~ one week of computer time per work unit (whatever 1 week is)and this was design. this is a big project. But it would be nice to see some clearer statistics including a sense of variation, not just averages.

"Unfortunately I doubt work units can be made any smaller as each encompass a single protein. When proteins fold each and every amino acid contributes to its characteristic folding pathway - therefore the whole peptide is required for the calculation. A change to a single amino acid in even the largest of proteins can render it inactive and alter its folding characteristics - the basis for many diseases. This has always been the problem in determining how proteins fold and their shear size requires such high computational requirements."

I get the point on how it will probably be hard (and require a LOT of work to divide the WU's up more at the get-go. But, there would be one easy solution to this.
As we all watch the percentage gauges from time to time to see if progress is being made - there seems to be an obvious way to break it up a little.
Have each computer work a set amount of % for each WU, then send in the partial results and treat that partial as one WU(Measuring 0- 100% within those 5% of the grand total).

I guess this could cause strain on those with poor connections (dialup and such... I'm told that they still exist somewhere, although I'm not sure where :) sooo... one could imagine a button that allowed you to just report the partial and continue onwards to the next 5% chip of the total WU, with a new stop at 10% and so on...