Hello Liu-- Sorry for the slow response, but I'm catching up after traveling for a while.
> > I am using PRE data to determine protein structure, but I don't > > understand how the Q factor is calculated. An detailed example is > > shown below. > > As shown below, the output file says the Q-factor for Cysp107hn is > > 0.944. However if I calculate the Q-factor using > > Q=sqrt(sum(diff^2)/sum(obs^2)), then Q is 0.4512. > If I calculate the Q-factor using > Q=sqrt(sum((obs-calc)^2)/sum(obs^2)), I also got the same value > (0.944). > But please note that I changed the type of penalty function into > "square", not the default "harmonic". In this case, 'diff' value is > not equal to 'calc'-'obs'. Clearly, the expression Q=sqrt(sum((obs-calc)^2)/sum(obs^2)) is always used. > For example, in the input file it gives a restrain as show below. > > assign (resid 78 and name HN) (resid 107 and name > NS1) 4080 4000 4000 > > This is a disappeared residue without precise R2, so I give a very > broad range 80-8080. In calculated structure, the calc R2 is 80.67, as > show below. > > id atoms in set i atoms in set j > obs calc diff weight S^2 S^2ang S^2rad > 4 ( 78 TYR HN)(ALT0 107 CYSP NS1) 4080 > 80.67 0 6.25e-08 1 1 1 > > As for this residue, 'obs'-'calc' is 3999.33, while 'diff' is 0. So if > Q is calculated using Q=sqrt(sum(diff^2)/sum(obs^2)), the value is > much smaller than Q=sqrt(sum((obs-calc)^2)/sum(obs^2). > One option might be to separate residues with missing gamma values into separate energy terms which uses a square potential type. The other terms should use a harmonic potential anyway. Then the Q-factors for the non-missing gamma values should be reported in a meaningful way. best regards-- Charles _______________________________________________ Xplor-nih mailing list Xplor-nih@cake.cit.nih.gov http://cake.cit.nih.gov/mailman/listinfo/xplor-nih
