Hi Tim, With small test sets, R-free doesn't become meaningless you just have to take into account that R-free has an error margin which is higher than for cases with a large test set. Few people report this error margin, but with a small data set you can easily do K-fold cross validation. I.e. do K refinements with K = 1/(test set fraction) and report R and R-free as averages with a standard deviation (instead of what we call cross validation, but is actually holdout validation). The CCP4 program freerflag already splits your data set in K groups to make it easier for the user. I do this automatically in PDB_REDO if the test set contains fewer than 500 reflections. It's amazing how much R-free is influenced by the choice of ones test set.
Cheers, Robbie > Date: Wed, 16 May 2012 16:06:24 +0200 > From: t...@shelx.uni-ac.gwdg.de > Subject: Re: [ccp4bb] correlations of B-factors and resolution > To: CCP4BB@JISCMAIL.AC.UK > > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Dear Qiang, > > without much explanation, rather from experience, the average B-factor > rises as resolution drops. It does make sense in a way because high > B-factors indicate some degree of disorder and disorder is usually the > cause for the resolution limit. 48A^2 for a 2.4A structure sound > perfectly fine with me, I would not worry provided that all other > statistices seem sound. > > High solvent content surely affects the B-values. The larger the > solvent channels and smaller the contact area between the molecules, > the more likely they become less stable and less ordered. > > R and Rfree seem also very good, although the gap is relatively tight. > Did you make sure your Rfree set contains at least 500 reflections? > The default of 5% often used, can lead to fewer reflections than 500 > at medium or low resolution, and with less than 500 reflection Rfree > becomes statistically meaningless - at least according to Axel > Brunger's article about that topic. > > Cheers, > Tim > > On 05/16/12 15:46, Qiang Chen wrote: > > Dear all, > > > > I have a 2.4A structure(pdb code 3LAF)with an average protein > > b-factor of 48. I wonder whether it's acceptable. Is there a direct > > correlation of b-factor and resolution? The R and Rfree are 21.1% > > and 23.1%, respectively. This structure has a very high solvent > > content, 75%. Does it affect the b-factors? > > > > Thanks a lot! > > > > Qiang > > > > > > The information in this e-mail is intended only for the person to > > whom it is addressed. If you believe this e-mail was sent to you in > > error and the e-mail contains patient information, please contact > > the Partners Compliance HelpLine at > > http://www.partners.org/complianceline . If the e-mail was sent to > > you in error but does not contain patient information, please > > contact the sender and properly dispose of the e-mail. > > > > - -- > - -- > Dr Tim Gruene > Institut fuer anorganische Chemie > Tammannstr. 4 > D-37077 Goettingen > > GPG Key ID = A46BEE1A > > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.12 (GNU/Linux) > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ > > iD8DBQFPs7RgUxlJ7aRr7hoRAnS8AJ472kwIWxf7rqDOhEPSBG5ipvQOWQCeNHNk > bum4yGTB56Wtt0JbkixleCw= > =uIfE > -----END PGP SIGNATURE-----
