On Sat, Mar 9, 2013 at 5:08 AM, Machius, Mischa Christian
I don't know how difficult the following would be to implement in your current system, so forgive me if it's too ambitious: Ideally, the user would be able to choose what to display. I am thinking of a column info system along the lines of iTunes where one can select which data should be shown (right-click on any column header). One should also be able to sort on any column (ascending/descending).
Things to show would include (nut are not limited to): R, Rfree, Rfree-R, variation of LLG and R factors in the last cycle compared to penultimate cycle (to test for convergence), Ramachandran outliers, rotamer outliers, clash score, input weights, LLG, map correlation coefficients, etc., all the goodies :)
Basically the problem right now is that almost all of these are stored in separate files, one per job - the R-free is shown in the job history because it's the only statistic stored centrally. In order to display these other statistics and sort them, I either need to significantly refactor the way the job data are stored, or read in the files for every job to extract the necessary information. I'm a little concerned about the overhead involved - the reason they're stored separately is to be able to load the basic job history as quickly as possible. I think there is a way to minimize this by caching some of the data, but it's not trivial, and will probably take me several tries to get it right. -Nat