Predicting pK_a for Small Molecules on Public and In‐house Datasets Using Fast Prediction Methods Combined with Data Fusion

Abstract

Data fusion approach was investigated in the context of pK a prediction for 391 small molecules derived from a public data source as well as for 681 compounds from an internal corporate database. Four different pKa prediction methods (Simulations Plus ADMET-Predictor S + pKa, ACD/Labs Percepta Classic, ACD/Labs Percepta GALAS and Epik) were used to predict the most acidic or basic pKa for each of the compounds. By using data fusion, the median absolute error for the internal compounds was reduced from the best performing single model's value of 0.69 down to 0.50. In addition to the improved accuracy, data fusion also enabled predictions for all of the compounds in the dataset as individual methods failed on some of the molecules.

Understanding OxymaPure as a Peptide Coupling Additive: A Guide to New Oxyma Derivatives

“…In this regard and to rationalize our previous results, and more importantly for the development of new ones, we have performed an in silico study using ACD/PhysChem Suite software and the p K a GALAS algorithm available in it to calculate the acid ionization constant values of various oximes and other additives (Table ). Like the p K a Classic method, which is a variation of a classical Hammett–Taft approach and is available as an alternative within the said software, the GALAS algorithm is based on analogous fundamental considerations. − However, instead of largely relying on equations and parameters quantified by other authors, it is developed entirely in-house by ACD/Labs, parameterized “from scratch” using an internal training set of >18 000 compounds with available experimental p K a measurement data. The custom nature of the p K a GALAS model allows for greater flexibility in using various ad hoc adjustments and modifications, going beyond the scope of the concepts considered in the classic Hammett–Taft approach where needed.…”

Curcumin – Bioavailability Enhancement by Prodrug Approach and Novel Formulations

“…Data fusion for prediction of pKa, in which the public data sources as well as an internal corporate database of small molecules are fused by using four different prediction models such as Simulations Plus, ADMET‐Predictor, ACD/Labs Percepta Classic, ACD/Labs Percepta GALAS and Epik. This new data fusion approach improves accuracy due to reduced median absolute errors of the internal compounds [121] . The permeability of curcumin in Caco‐ 2 cell monolayers and the poor absorption of curcumin through the gut lumen were reported due to degradation in the HBSS buffer across the Caco‐2 cell model during transportation [122] …”

“…This new data fusion approach improves accuracy due to reduced median absolute errors of the internal compounds. [121] The permeability of curcumin in Caco-2 cell monolayers and the poor absorption of curcumin through the gut lumen were reported due to degradation in the HBSS buffer across the Caco-2 cell model during transportation. [122] The energy distribution pattern of the curcumin was predicted by using Gaussview by using B3LYP/6-311 + + (d, p) model.…”

Drug-like properties of serial phenanthroindolizidine alkaloid compounds: ADMET characteristic prediction and validation

“…Thus, to understand the drug-like properties exhibited by a large series of molecules, the easiest and least expensive method involves prediction in silico, and the best compounds can be chosen as candidates and compounds with poor drug-like properties can be removed [18,19]. Thereafter, prediction software can also exhibit large deviations and interfere with reaching valid conclusions [20]. Thus, the best candidates need to be accurately measured before entering rigorous preclinical research [21].…”