Marina Gorostiola González

Building reliable and robust quantitative structure–property relationship (QSPR) models is a challenging task. First, the experimental data needs to be obtained, analyzed and curated. Second, the number of available methods is continuously growing and evaluating different algorithms and methodologies can be arduous. Finally, the last hurdle that researchers face is to ensure the reproducibility of their models and facilitate their transferability into practice. In this work, we introduce QSPRpred, a toolkit for analysis of bioactivity data sets and QSPR modelling, which attempts to address the aforementioned challenges. QSPRpred’s modular Python API enables users to intuitively describe different parts of a modelling workflow using a plethora of pre-implemented components, but also integrates customized implementations in a “plug-and-play” manner. QSPRpred data sets and models are directly serializable, which means they can be readily reproduced and put into operation after training as the models are saved with all required data pre-processing steps to make predictions on new compounds directly from SMILES strings. The general-purpose character of QSPRpred is also demonstrated by inclusion of support for multi-task and proteochemometric modelling. The package is extensively documented and comes with a large collection of tutorials to help new users. In this paper, we describe all of QSPRpred’s functionalities and also conduct a small benchmarking case study to illustrate how different components can be leveraged to compare a diverse set of models. QSPRpred is fully open-source and available at https://github.com/CDDLeiden/QSPRpred.Scientific ContributionQSPRpred aims to provide a complex, but comprehensive Python API to conduct all tasks encountered in QSPR modelling from data preparation and analysis to model creation and model deployment. In contrast to similar packages, QSPRpred offers a wider and more exhaustive range of capabilities and integrations with many popular packages that also go beyond QSPR modelling. A significant contribution of QSPRpred is also in its automated and highly standardized serialization scheme, which significantly improves reproducibility and transferability of models.

Personalized oncology has revolutionized cancer treatment by targeting specific genetic aberrations in tumors. However, the identification of suitable targets for anti-cancer therapies remains a challenge. In this study, we introduce a knowledge graph approach to prioritize cancer mutations with clinical, functional, and structural significance as potential therapeutic targets. Focusing on the human kinome, we integrate protein-protein interaction and patient-centric mutation networks to construct a comprehensive network enriched with transcriptomic, structural, and drug response data, together covering five layers of information. Moreover, we make the constructed knowledge graph publicly available, along with a plethora of scripts to facilitate further annotation and expansion of the network. Interactive visualization resources are also provided, ensuring accessibility for researchers regardless of computational expertise and enabling detailed analysis by cancer type and individual layers of information. This comprehensive resource has the potential to identify relevant mutations for targeted therapeutic interventions, thereby advancing personalized oncology and improving patient outcomes.

Bioactivity prediction is essential in computational drug discovery, particularly within virtual screening campaigns. Despite advancements in model architectures and features, the sparsity and quality of relevant training data remain a major bottleneck. Notably, genetic variance annotation, crucial for understanding variant-specific bioactivity, is often neglected. Key efforts to tackle these issues are conducted by public bioactivity databases such as ChEMBL, but these are not free of challenges. Here, a comprehensive analysis of the extent and distribution of bioactivity data tested on genetic variants across organisms, protein families, individual targets, and specific variants, for the first time characterises in detail the genetic variability landscape in the ChEMBL database and sheds light on the range and consequences of protein amino acid substitutions in bioactivity data distribution and modelling. Furthermore, an extensive set of analysis resources (Python package and notebooks) and a variant-annotated bioactivity dataset are made available to help replicate the analyses described here for any protein of interest and make informed decisions regarding the quality of data for modelling. Finally, the potential to extract variants and subsets of the chemical space with desirable inter-variant bioactivity profiles is demonstrated for data-rich proteins. This approach contributes to more reliable bioactivity modelling, aids noise reduction and informs decision-making in computational drug discovery.

The human equilibrative nucleoside transporter 1 (SLC29A1, hENT1) is a solute carrier that modulates the passive transport of nucleosides and nucleobases, such as adenosine. This nucleoside regulates various physiological processes, such as vasodilation and -constriction, neurotransmission and immune defense. Marketed drugs such as dilazep and dipyridamole have proven useful in cardiovascular afflictions, but the application of hENT1 inhibitors can be beneficial in a number of other diseases. In this study, 39 derivatives of dilazep’s close analogue ST7092 were designed, synthesized and subsequently assessed using [³H]NBTI displacement assays and molecular docking. Different substitution patterns of the trimethoxy benzoates of ST7092 reduced interactions within the binding pocket, resulting in diminished hENT1 affinity. Conversely, [³H]NBTI displacement by potentially covalent compounds 14b, 14c, and 14d resulted in high affinities (K_i values between 1.1 and 17.5 nM) for the transporter, primarily by the ability of accommodating the inhibitors in various ways in the binding pocket. However, any indication of covalent binding with amino acid residue C439 remained absent, conceivably as a result of decreased nucleophilic residue reactivity. In conclusion, this research introduces novel dilazep derivatives that are active as hENT1 inhibitors, along with the first high affinity dilazep derivatives equipped with an electrophilic warhead. These findings will aid the rational and structure-based development of novel hENT1 inhibitors and pharmacological tools to study hENT1's function, binding mechanisms, and its relevance in (patho)physiological conditions.

Cancer remains a leading cause of mortality worldwide and calls for novel therapeutic targets. Membrane proteins are key players in various cancer types but present unique challenges compared to soluble proteins. The advent of computational drug discovery tools offers a promising approach to address these challenges, allowing for the prioritization of “wet-lab” experiments. In this review, we explore the applications of computational approaches in membrane protein oncological characterization, particularly focusing on three prominent membrane protein families: receptor tyrosine kinases (RTKs), G protein-coupled receptors (GPCRs), and solute carrier proteins (SLCs). We chose these families due to their varying levels of understanding and research data availability, which leads to distinct challenges and opportunities for computational analysis. We discuss the utilization of multi-omics data, machine learning, and structure-based methods to investigate aberrant protein functionalities associated with cancer progression within each family. Moreover, we highlight the importance of considering the broader cellular context and, in particular, cross-talk between proteins. Despite existing challenges, computational tools hold promise in dissecting membrane protein dysregulation in cancer. With advancing computational capabilities and data resources, these tools are poised to play a pivotal role in identifying and prioritizing membrane proteins as personalized anticancer targets.

Building reliable and robust quantitative structure-property relationship (QSPR) models is a challenging task. First, the experimental data needs to be obtained, analyzed and curated. Second, the number of available methods is continuously growing and evaluating different algorithms and methodologies can be arduous. Finally, the last hurdle that researchers face is to ensure the reproducibility of their models and facilitate their transferability into practice. In this work, we introduce QSPRpred, a toolkit for analysis of bioactivity data sets and QSPR modelling, which attempts to address the aforementioned challenges. QSPRpred's modular Python API enables users to intuitively describe different parts of a modelling workflow using a plethora of pre-implemented components, but also integrate customized implementations in a "plug-and-play" manner. QSPRpred data sets and models are directly serializable, which means they can be readily reproduced and put into operation after training as the models are saved with all required data pre-processing steps to make predictions on new compounds directly from SMILES strings. The general-purpose character of QSPRpred is also demonstrated by inclusion of support for multi-task and proteochemometric modelling. The package is extensively documented and comes with a large collection of tutorials to help new users. In this paper, we describe all of QSPRpred's functionalities and also conduct a small benchmarking case study to illustrate how different components can be leveraged to compare a diverse set of models. QSPRpred is fully open-source and available at https://github.com/CDDLeiden/QSPRpred. Scientific Contribution QSPRpred aims to provide a complex, but comprehensive Python API to conduct all tasks encountered in QSPR modelling from data preparation and analysis to model creation and model deployment. In contrast to similar packages, QSPRpred offers a wider and more exhaustive range of capabilities and integrations with many popular packages that also go beyond QSPR modelling. A significant contribution of QSPRpred is also in its automated and highly standardized serialization scheme, which significantly improves reproducibility and transferability of models.

Developments in computational omics technologies have provided new means to access the hidden diversity of natural products, unearthing new potential for drug discovery. In parallel, artificial intelligence approaches such as machine learning have led to exciting developments in the computational drug design field, facilitating biological activity prediction and de novo drug design for molecular targets of interest. Here, we describe current and future synergies between these developments to effectively identify drug candidates from the plethora of molecules produced by nature. We also discuss how to address key challenges in realizing the potential of these synergies, such as the need for high-quality datasets to train deep learning algorithms and appropriate strategies for algorithm validation.

Proteochemometric (PCM) modelling is a powerful computational drug discovery tool used in bioactivity prediction of potential drug candidates relying on both chemical and protein information. In PCM features are computed to describe small molecules and proteins, which directly impact the quality of the predictive models. State-of-the-art protein descriptors, however, are calculated from the protein sequence and neglect the dynamic nature of proteins. This dynamic nature can be computationally simulated with molecular dynamics (MD). Here, novel 3D dynamic protein descriptors (3DDPDs) were designed to be applied in bioactivity prediction tasks with PCM models. As a test case publicly available G protein-coupled receptor (GPCR) MD data from GPCRmd was used. GPCRs are membrane-bound proteins, which are activated by hormones and neurotransmitters, and constitute an important target family for drug discovery. GPCRs exist in different conformational states that allow transmission of diverse signals and that can be modified by ligand interactions, among other factors. To translate the MD-encoded protein dynamics two types of 3DDPDs were considered: one-hot encoded residue-specific (rs) and embedding-like protein-specific (ps) 3DDPDs. The descriptors were developed by calculating distributions of trajectory coordinates and partial charges, applying dimensionality reduction, and subsequently condensing them into vectors per residue or protein, respectively. 3DDPDs were benchmarked on a number of PCM tasks against state-of-the-art non-dynamic protein descriptors. Our rs- and ps3DDPDs outperformed non-dynamic descriptors in regression tasks using a temporal split, and showed comparable performance with a random split and in all classification tasks. Combinations of non-dynamic descriptors with 3DDPDs did not result in increased performance. Finally, the power of 3DDPDs to capture dynamic fluctuations in mutant GPCRs was explored. The results presented here show the potential of including protein dynamic information on machine learning tasks, specifically bioactivity prediction, and open opportunities for applications in drug discovery, including oncology.

G Protein-coupled receptors (GPCRs) are the most frequently exploited drug target family, moreover they are often found mutated in cancer. Here we used a dataset of mutations found in patient samples derived from the Genomic Data Commons and compared it to the natural human variance as exemplified by data from the 1000 genomes project. We explored cancer-related mutation patterns in all GPCR classes combined and individually. While the location of the mutations across the protein domains did not differ significantly in the two datasets, a mutation enrichment in cancer patients was observed among class-specific conserved motifs in GPCRs such as the Class A “DRY” motif. A Two-Entropy Analysis confirmed the correlation between residue conservation and cancer-related mutation frequency. We subsequently created a ranking of high scoring GPCRs, using a multi-objective approach (Pareto Front Ranking). Our approach was confirmed by re-discovery of established cancer targets such as the LPA and mGlu receptor families, but also discovered novel GPCRs which had not been linked to cancer before such as the P2Y Receptor 10 (P2RY10). Overall, this study presents a list of GPCRs that are amenable to experimental follow up to elucidate their role in cancer.