MF-SuP-pKa: multi-fidelity modeling with subgraph pooling mechanism for pKa prediction



Acid-base dissociation constant (pKa) is a key physicochemical parameter in chemical science, especially in organic synthesis and drug discovery. Current methodologies for pKa prediction still suffer from limited applicability domain and lack of chemical insight. Here we present MF-SuP-pKa (Multi-Fidelity modeling with Subgraph Pooling for pKa prediction), a novel pKa prediction model that utilizes subgraph pooling, multi-fidelity learning and data augmentation. In our model, a knowledge-aware subgraph pooling strategy was designed to capture the local and global environments around the ionization sites for micro-pKa prediction. To overcome the scarcity of accurate pKa data, low-fidelity data (computational pKa) was used to fit the high-fidelity data (experimental pKa) through transfer learning. Moreover, we implemented knowledge-guided data augmentation on the pre-training data according to the consistency between acidic pKa and basic pKa. The final MF-SuP-pKa model was constructed by pre-training on the augmented ChEMBL data set and fine-tuning on the DataWarrior data set. The ablation results prove that MF-SuP-pKa gains essential benefits from subgraph pooling, multi-fidelity learning, and data augmentation. Extensive evaluation on the DataWarrior data set and three benchmark data sets shows that MF-SuP-pKa achieves superior performances to the state-of-the-art pKa prediction models while requires much less high-fidelity training data. Compared with Attentive FP, MF-SuP-pKa achieves 23.83% and 20.12% improvement in terms of mean absolute error (MAE) on the acidic and basic sets, respectively.


Supplementary material

Supplementary materials of MF-SuP-pKa
Table S1. The initial atom and bond features for graph-based methods. Table S2. The performance of different machine learning algorithms on the DataWarrior data set. Table S3. The hyperparameters of each model. Table S4. The performance of MF-SuP-pKa on the external test set. Figure S1. Visualization of the ionizable atom labels on 20 representative amphoteric molecules. Figure S2. Distribution of the pairwise Tanimoto similarities between the DataWarrior data set and the external test set. Figure S3. Results of micro-pKa prediction on the SAMPL6 data set using MF-SuP-pKa. Figure S4. Distribution of molecular size for the DataWarrior data set.