The R value is used to assess progress in the refinement of a model from X-ray crystallographic data, and can be used as one factor in evaluating the quality of a model. R is a measure of error between the observed intensities from the diffraction pattern and the predicted intensities that are calculated from the model. R values of 0.20 or less are taken as evidence that the model is reliable.
As a rule of thumb, models with R values substantially exceeding (resolution/10) should be treated with caution. Thus, if the resolution of a model is 2.5 Å, that model's R value should not exceed 0.25. Completely erroneous models (e.g. random models) give R values of 0.40 to 0.60.
However, R values themselves must be treated with caution. Unlike the Free R, acceptable R values can be achieved despite serious errors in the model, as demonstrated unequivocally by Kleywegt & Brünger. In fact analysis has shown that this value is frequently underestimated so that a final model is not as good as it should be. A different statistical value, CC*, has been proposed to assess model and data quality on the same scale and reveal when data quality is limiting model improvement .
One famous pitfall that can result in a misleading R value is the addition of substantially more than one water molecule per amino acid.
- ↑ Kleywegt GJ, Brunger AT. Checking your imagination: applications of the free R value. Structure. 1996 Aug 15;4(8):897-904. PMID:8805582
- ↑ 2.0 2.1 Karplus PA, Diederichs K. Linking crystallographic model and data quality. Science. 2012 May 25;336(6084):1030-3. PMID:22628654 doi:10.1126/science.1218231
- ↑ Evans P. Biochemistry. Resolving some old problems in protein crystallography. Science. 2012 May 25;336(6084):986-7. PMID:22628641 doi:10.1126/science.1222162