Evaluation of Log P, pKa and Log D Predictions from the SAMPL7 Blind Challenge

Abstract

<div>The Statistical Assessment of Modeling of Proteins and Ligands (SAMPL) challenges focuses the computational modeling community on areas in need of improvement for rational drug design. The SAMPL7 physical property challenge dealt with prediction of octanol-water partition coefficients and pKa for 22 compounds. </div><div>The dataset was composed of a series of N-acylsulfonamides and related bioisosteres.</div><div>17 research groups participated in the logP challenge, submitting 33 blind submissions total. For the pKa challenge, 7 different groups participated, submitting 9 blind submissions in total. Overall, the accuracy of octanol-water logP predictions in the SAMPL7 challenge was lower than octanol-water logP predictions in SAMPL6, likely due to a more diverse dataset. Compared to the SAMPL6 pKa challenge, accuracy remains unchanged in SAMPL7.</div><div>Interestingly, here, though macroscopic pKa values were often predicted with reasonable accuracy, there was dramatically more disagreement among participants as to which microscopic transitions produced these values (with methods often disagreeing even as to the sign of the free energy change associated with certain transitions), indicating far more work needs to be done on pKa prediction methods.</div>

Prediction of n-octanol/water partition coefficients and acidity constants (pKa) in the SAMPL7 blind challenge with the IEFPCM-MST model

“…The predicted log P values are listed in Table 1. The rootmean square deviation (rmsd) between IEFPCM/MST results and experimental data is 1.03 log units, which places our results among the most accurate values in the comparison with both physical (rank 2nd) and global (comprising all submissions within empirical and physical categories; rank 8th) methods [21], taking into account the small differences observed between methods with rmsd ≤ 1 (see Supporting Information Fig. S1).…”

“…Here, we report the results obtained for predicting the log P and pK a for a group of sulfonamide-containing compounds. The results are discussed in light of the experimental data provided by the organizers of SAMPL7 [21] and the theoretical estimates reported by others groups, as well as with the IEFPCM/MST results obtained in previous editions of this contest [22,23].…”

“…Overall, the results support the suitability of our QMbased approach for computing log P and pK a properties. SAMPL6 blind challenge mainly relied on rigid compounds [38], but SAMPL7 presented more complex compounds considering both chemical diversity and flexibility [21]. In the blind challenges mentioned above, the Frog tool has been used to explore the conformational space in our QM workflow mainly due to the good balance between computational cost and accuracy of the conformer ensemble [25].…”

Evaluation of log P, pKa, and log D predictions from the SAMPL7 blind challenge

“…Submission ClassicalGSG DB3 is an empirical method that employed neural networks (NNs) where the inputs are molecular features generated using a method called Geometric Scattering for Graphs (GSG) [84][85][86]. In GSG, atomic features are transformed into molecular features using the graph molecular structure.…”

Predicting partition coefficients for the SAMPL7 physical property challenge using the ClassicalGSG method

“…To help meet this need, the statistical assessment of the modeling of proteins and ligands (SAMPL) [ 25 ] project recently created a distinct blind challenge for predicting log P allowing fair evaluation and comparison of different log P prediction methods (SAMPL6 in 2019 [ 26 ] and SAMPL7 in 2020 [ 27 ]). In this challenge, the participants predict log P for a set of drug-like molecules and the predictions are assessed using experimental log P values that are revealed later.…”