We therefore propose that for compounds with a molecular weight range corresponding to common poorly soluble drugs, properties relating to molecular size is the dominating factor determining glass-forming ability, whereas for limited series of compounds with similar molecular weight, the Tg,red may be more useful for predictions. Some publications highlight the role of the configurational entropy difference between the amorphous and crystalline state, and that compounds with higher Mw have more complex molecular structure and hence, are less likely to exist in an ordered crystalline state ( Bhugra and Pikal, 2008, Graeser et al., 2009 and Zhou

et al., 2002). Therefore, there seems to be a rational behind using the Mw as an easily obtained surrogate for description of configurational entropy, although the latter property BKM120 also is dependent on other structural features, e.g. number of rotatable bonds. Further, it has been suggested that the complexity associated with larger molecules means that it has to probe a larger number of possible conformations and configurations to find an ordered (crystalline) packing structure during solidification ( Bhugra and Pikal, 2008). It is

appealing to imagine the tendency of becoming either amorphous or crystalline as being dependent on the molecular process of probing the various possible conformations and configurations (related the configurational space, and hence to the Mw of the compound) and the time available to find a configuration that will produce an ordered crystal unit during Anti-diabetic Compound Library solidification (related to the Tg,red at constant

cooling conditions). In the present study, the dominating factor for glass-formation seems to be Mw. In Fig. 2 the relation between Mw and glass-forming ability is visualized. From our analysis, based on a large structurally diverse dataset we suggest that compounds with Mw above 300 g/mole are likely to be transformed to the corresponding glass using standard production/amorphization technologies, whereas compounds with Mw below this value will be difficult to produce amorphous. It should be kept in mind that we base this conclusion on compounds having a melting point higher than 140 °C. However, the general applicability of this rule-of-thumb was confirmed by applying the analysis Cytidine deaminase on the 51 compounds studied by Baird et al. (2010). For this dataset, 84% of the compounds were correctly sorted with regard to their glass-forming ability when using Mw of 300 g/mole as the cut-off value. In the same way as for glass-forming ability, the glass stability was analysed step-wise. The thermodynamic properties did, again, not result in a significant model for dry stability. The variable selection after including the Tg-related properties to the model development resulted in that Tg was found to be the single most important property, and did by itself predict 65% of the compounds accurately ( Fig. 3A).