[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Luzatti plot

***  For details on how to be removed from this list visit the  ***
***    CCP4 home page http://www.dl.ac.uk/CCP/CCP4/main.html    ***

(Perhaps a tad late, but I haven't read my mail in 3 weeks.)

I have recently done a study of the relationship between data
quality, model quality, and accuracy (still need to properly
analyse and write up the results).  As for coordinate error
estimates, I looked at cv-Luzzati and cv-SIGMAA from CNS, and
min and max estimated error and DPI from SFCHECK.  Of these
five statistics, the *only* one that correlates somewhat
meaningfully with model accuracy is the "maximum estimated
error" as calculated by SFCHECK (and then *only* for models
that are close to finished, i.e. not for your first MIR,
MAD or MR model).  [All refinements used CNS and its MLF

[This is just one of dozens of findings that range from
"I always suspected that" to "What ??? I don't believe it !"
Look forward to more tidbits as relevant discussion
subjects come up on this bulletin board ;-)]