QSAR-derived affinity fingerprints (part 2): modeling performance for potency prediction

Abstract

Affinity fingerprints report the activity of small molecules across a set of assays, and thus permit to gather information about the bioactivities of structurally dissimilar compounds, where models based on chemical structure alone are often limited, and model complex biological endpoints, such as human toxicity and in vitro cancer cell line sensitivity. Here, we propose to model in vitro compound activity using computationally predicted bioactivity profiles as compound descriptors. To this aim, we apply and validate a framework for the calculation of QSAR-derived affinity fingerprints (QAFFP) using a set of 1360 QSAR models generated using K i , K d , IC 50 and EC 50 data from ChEMBL database. QAFFP thus represent a method to encode and relate compounds on the basis of their similarity in bioactivity space. To benchmark the predictive power of QAFFP we assembled IC 50 data from ChEMBL database for 18 diverse cancer cell lines widely used in preclinical drug discovery, and 25 diverse protein target data sets. This study complements part 1 where the performance of QAFFP in similarity searching, scaffold hopping, and bioactivity classification is evaluated. Despite being inherently noisy, we show that using QAFFP as descriptors leads to errors in prediction on the test set in the ~ 0.65-0.95 pIC 50 units range, which are comparable to the estimated uncertainty of bioactivity data in ChEMBL (0.76-1.00 pIC 50 units). We find that the predictive power of QAFFP is slightly worse than that of Morgan2 fingerprints and 1D and 2D physicochemical descriptors, with an effect size in the 0.02-0.08 pIC 50 units range. Including QSAR models with low predictive power in the generation of QAFFP does not lead to improved predictive power. Given that the QSAR models we used to compute the QAFFP were selected on the basis of data availability alone, we anticipate better modeling results for QAFFP generated using more diverse and biologically meaningful targets. Data sets and Python code are publicly available at https ://githu b.com/isidr oc/QAFFP _regre ssion .

QSAR-derived affinity fingerprints (part 1): fingerprint construction and modeling performance for similarity searching, bioactivity classification and scaffold hopping

“…The QAFFP fingerprint was constructed using the predictions of high quality Random Forest models trained on freely available (i.e., non-proprietary) data covering diverse sets of molecular targets. Its performance was compared with that of the Morgan2 fingerprint (i.e., Morgan fingerprint with radius 2, the RDKit [52] implementation of the widely-used ECFP4 fingerprint [2]) for similarity searching, for the classification of compounds as active or inactive and for scaffold hopping. In addition to similarity searching, compound classification and scaffold hopping, QAFFP fingerprint was also applied in regression setting to predict compound in vitro potency, as described in the accompanying paper [53].…”

Profiling and analysis of chemical compounds using pointwise mutual information

“…with the aim to improve the interpretability of the resulting interrelation profiles. Later, we will also investigate the utility of hybrid feature vectors containing interrelation profiles concatenated with, for example, QAFFP biological fingerprints [63,64] or with other features of interest. We plan to use interrelation profiling in various cheminformatics applications, such as in biological activity classification or potency prediction, focused chemical library construction, diversity data selection or ensemble modeling using RFT together with domain-specific models for, e.g., natural product likeness assessment [65][66][67].…”

Profiling and analysis of chemical compounds using pointwise mutual information

“…with the aim to improve the interpretability of the resulting interrelation profiles. Later, we will also investigate the utility of hybrid feature vectors containing interrelation profiles concatenated with, for example, QAFFP biological fingerprints [64,65] or with other features of interest. We plan to use interrelation profiling in various cheminformatics applications, such as in biological activity classification or potency prediction, focused chemical library construction, diversity data selection or ensemble modeling using RFT together with domain-specific models for, e.g., natural product likeness assessment [66][67][68].…”