GRIND Limitations and Problems

The results of any QSAR model (not only GRIND generated models) express the correlation between the differences in the structure of the compounds and their differences in biological properties. Structural features which are common to every compound in the series are simply not considered in the analysis. Any pharmacophoric interpretation of the results of a QSAR analysis will miss structural features shared by all the compounds. This fact is not trivial, because most series are constituted mainly of compounds with a certain degree of affinity for the receptor and this reveals that most of them have relevant common features. Therefore, as stated previously, none of these features will be present in the results of the models and no pharmacophore derived from this analysis will incorporate them. Moreover, the QSAR model results also inform us of structural features that are detrimental for the activity, while the pharmacophores are often focused only on the structural features that are needed to obtain active compounds. 

Another reason to avoid a pharmacophoric interpretation of the 3D QSAR/ GRIND results is the presence of correlation effects. Often, the compounds in the series studied belong to different structural families, each one characterised by a number of structural features, like the presence of certain groups, rings or chains. 

The problem is that all these features are present or absent together. For this reason, the results of the QSAR analysis will assign the same importance to all these features and if only one of them is relevant for the activity, no model would be able to distinguish this single feature from the others, since all are present or absent simultaneously in the tested compounds. However, if these results are extrapolated to external compounds, in which maybe only some of the features are present, the predictions will be wrong. 

For all the above reasons, the results of the QSAR models obtained with GRIND should not be considered as pharmacophores. A correct and sensible interpretation of these results would be extremely useful, but its over-interpretation can be misleading and produce unrealistic expectations. It should also be stressed that the above-mentioned considerations are applicable to most QSAR and 3D QSAR results and are not a problem strictly linked to GRIND. 

---