Rge calculationWe calculated QM atomic charges for all the combinations of QM theory level, basis set and population analysis for which we’ve got EEM parameters (see Table 1). Specifically, atomic charges had been calculated for these eight QM approaches: HF/STO3G/MPA, HF/61G/MK, and B3LYP/61G with MPA, NPA, Hirshfeld, MK, CHELPG and AIM). The QM charge calculations have been carried out utilizing Gaussian09 [54]. Inside the case of AIM population evaluation, the output from Gaussian09 was further processed by the software program package AIMAll [55].Data set for phenolsAn aspect which can be crucial for the applicability of the pKa prediction strategy is its transferability to other chemical classes. In this operate, we present a case study showing the efficiency in the approach on carboxylic acids, which are also really typical testing molecules for pKa prediction approaches [1921,43]. The data set includes 71 distinct molecules of carboxylic acids and their dissociated forms.94928-86-6 web The 3D structures of those molecules had been obtained in the very same way as for the phenols. The list of the molecules, like their names, NCS numbers, CAS numbers and experimental pKa values can be located within the (Further file 3: Table S1b). The SDF files using the 3D structures from the molecules and their dissociated types are also incorporated in the (More file 2: Molecules).pKa valuesThere are two principal approaches to make a QSPR model to get a feature to become predicted. The initial will be to produce as generalThe experimental pKa values have been taken from the Physprop database [61].42225-04-7 site SvobodovVaekovet al.PMID:33529090 Journal of Cheminformatics 2013, 5:18 a r a http://www.jcheminf.com/content/5/Page 4 ofTable 1 Summary information regarding the EEM parameter sets employed in the present studyQM theory level basis set HF/STO3G MPA PA EEM parameter set name Svob2007 cbeg2 Svob2007 cmet2 Svob2007 chal2 Svob2007 hm2 Baek1991 Mort1986 HF/61G B3LYP/61G MK MPA NPA Jir2008 hf Chaves2006 Bult2002 mul Ouy2009 Ouy2009 elem Ouy2009 elemF Bult2002 npa Hir. MK CHELPG AIM Bult2002 hir Jir2008 mk Bult2002 mk Bult2002 che Bult2004 aim Svobodova et al. [44] Svobodova et al. [44] Svobodova et al. [44] Svobodova et al. [44] Baekelandt et al. [45] Mortier et al. [39] Jirouskova et al. [46] Chaves et al. [47] Bultinck et al. [48] Ouyang et al. [49] Ouyang et al. [49] Ouyang et al. [49] Bultinck et al. [48] Bultinck et al. [48] Jirouskova et al. [46] Bultinck et al. [48] Bultinck et al. [48] Bultinck et al. [50] Published by Year of publication 2007 2007 2007 2007 1991 1986 2008 2006 2002 2009 2009 2009 2002 2002 2008 2002 2002 2004 C, O, N, H, S C, O, N, H, S, Fe, Zn C, O, N, H, S, Br, Cl, F, I C, O, N, H, S, F, Cl, Br, I, Fe, Zn C, O, N, H, P, Al, Si C, O, N, H C, O, N, H, S, F, Cl, Br, I, Zn C, O, N, H, F C, O, N, H, F C, O, N, H, F C, O, N, H, F C, O, N, H, F C, O, N, H, F C, O, N, H, F C, O, N, H, S, F, Cl, Br, I, Zn C, O, N, H, F C, O, N, H, F C, O, N, H, F Components includedDescriptors and QSPR models for phenolsOur descriptors were atomic charges. We analyzed two types of QSPR models, namely QSPR models with 3 descriptors (3d QSPR models) and QSPR models with 5 descriptors (5d QSPR models). The 3d QSPR models made use of those descriptors which proved to be the most relevant for pKa prediction in our earlier study [24]. Consequently these descriptors were the atomic charge in the hydrogen atom from the phenolic OH group (qH ), the charge around the oxygen atom in the phenolic OH group (qO ), as well as the charge around the carbon atom binding the phenolic OH gr.