The influence of solid state information and descriptor selection on statistical models of temperature dependent aqueous solubility

Table 1 QSPR ready datasets

Dataset name	Instance identifier^a	Extra filtering?	No. instances	Total no. descriptors^b	Endpoint	Source^c	Integrated with crystal structures?
Avdeef_ExDPs_CS_False	[name]_[CAS no.]_[polymorph description]	Low quality data points removed [17]	364	4776	Enthalpy of solution	Avdeef [17] derived dataset	No
Avdeef_ExDPs_Cal_CS_False	[name]_[CAS no.]_[polymorph description]	Only calorimetry data points retained [17]	50	4776	Enthalpy of solution	Avdeef [17] derived dataset
Klimenko_CS_False	[name]_[CAS no.]_[polymorph description]_[temperature value]	No	882	3764	Solubility at some defined temperature	Klimenko et al. [14] derived dataset
Avdeef_ExDPs_CS_True	[name]_[CAS no.]_[refcode]	Low quality data points removed [17]	169	4820	Enthalpy of solution	Avdeef [17] derived dataset	Yes
Avdeef_ExDPs_Cal_CS_True	[name]_[CAS no.]_[refcode]	Only calorimetry data points retained [17]	30	4820	Enthalpy of solution	Avdeef [17] derived dataset
Klimenko_CS_True	[name]_[CAS no.]_[refcode]_[temperature value]	No	530	3808	Solubility at some defined temperature	Klimenko et al. [14] derived dataset

^aThe [name] (or [CAS no.]) and/or [polymorph description] could be “none”, denoting the absence of the relevant information
^bThis denotes the complete set of all 2D molecular, temperature (for the solubility datasets), melting point and, for crystal structure integrated datasets, lattice energy and 3D descriptors calculated for instances in this dataset. Different subsets were considered for different models, as described under “Descriptor combinations investigated”
^cSee Fig. 1

ISSN: 1758-2946