Georgios Drakakis

Diversity Selection of Compounds Based on ‘Protein Affinity Fingerprints’ Improves Sampling of Bioactive Chemical Space

Diversity selection is a frequently applied strategy for assembling high‐throughput screening libraries, making the assumption that a diverse compound set increases chances of finding bioactive molecules. Based on previous work on experimental ‘affinity fingerprints’, in this study, a novel diversity selection method is benchmarked that utilizes predicted bioactivity profiles as descriptors. Compounds were selected based on their predicted activity against half of the targets (training set), and diversity was assessed based on coverage of the remaining (test set) targets. Simultaneously, fingerprint‐based diversity selection was performed. An original version of the method exhibited on average 5% and an improved version on average 10% increase in target space coverage compared with the fingerprint‐based methods. As a typical case, bioactivity‐based selection of 231 compounds (2%) from a particular data set (‘Cutoff‐40’) resulted in 47.0% and 50.1% coverage, while fingerprint‐based selection only achieved 38.4% target coverage for the same subset size. In conclusion, the novel bioactivity‐based selection method outperformed the fingerprint‐based method in sampling bioactive chemical space on the data sets considered. The structures retrieved were structurally more acceptable to medicinal chemists while at the same time being more lipophilic, hence bioactivity‐based diversity selection of compounds would best be combined with physicochemical property filters in practice.

Extensions to In Silico Bioactivity Predictions Using Pathway Annotations and Differential Pharmacology Analysis: Application to Xenopus laevis Phenotypic Readouts

The simultaneous increase of computational power and the availability of chemical and biological data have contributed to the recent popularity of in silico bioactivity prediction algorithms. Such methods are commonly used to infer the ‘Mechanism of Action’ of small molecules and they can also be employed in cases where full bioactivity profiles have not been established experimentally. However, protein target predictions by themselves do not necessarily capture information about the effect of a compound on a biological system, and hence merging their output with a systems biology approach can help to better understand the complex network modulation which leads to a particular phenotype. In this work, we review approaches and applications of target prediction, as well as their shortcomings, and demonstrate two extensions of this concept which are exemplified using phenotypic readouts from a chemical genetic screen in Xenopus laevis. In particular, the experimental observations are linked to their predicted bioactivity profiles. Predicted targets are annotated with pathways, which lead to further biological insight. Moreover, we subject the prediction to further machine learning algorithms, namely decision trees, to capture the differential pharmacology of ligand‐target interactions in biological systems. Both methodologies hence provide new insight into understanding the Mechanism of Action of compound activities from phenotypic screens.

Extending in silico mechanism-of- action analysis by annotating targets with pathways: application to cellular cytotoxicity readouts

BACKGROUND An in silico mechanism-of-action analysis protocol was developed, comprising molecule bioactivity profiling, annotation of predicted targets with pathways and calculation of enrichment factors to highlight targets and pathways more likely to be implicated in the studied phenotype. RESULTS The method was applied to a cytotoxicity phenotypic endpoint, with enriched targets/pathways found to be statistically significant when compared with 100 random datasets. Application on a smaller apoptotic set (10 molecules) did not allowed to obtain statistically relevant results, suggesting that the protocol requires modification such as analysis of the most frequently predicted targets/annotated pathways. CONCLUSION Pathway annotations improved the mechanism-of-action information gained by target prediction alone, allowing a better interpretation of the predictions and providing better mapping of targets onto pathways.

Comparing global and local likelihood score thresholds in multiclass laplacian-modified Naive Bayes protein target prediction.

The increase of publicly available bioactivity data has led to the extensive development and usage of in silico bioactivity prediction algorithms. A particularly popular approach for such analyses is the multiclass Naïve Bayes, whose output is commonly processed by applying empirically-derived likelihood score thresholds. In this work, we describe a systematic way for deriving score cut-offs on a per-protein target basis and compare their performance with global thresholds on a large scale using both 5-fold cross-validation (ChEMBL 14, 189k ligand-protein pairs over 477 protein targets) and external validation (WOMBAT, 63k pairs, 421 targets). The individual protein target cut-offs derived were compared to global cut-offs ranging from -10 to 40 in score bouts of 2.5. The results indicate that individual thresholds had equal or better performance in all comparisons with global thresholds, ranging from 95% of protein targets to 57.96%. It is shown that local thresholds behave differently for particular families of targets (CYPs, GPCRs, Kinases and TFs). Furthermore, we demonstrate the discrepancy in performance when we move away from the training dataset chemical space, using Tanimoto similarity as a metric (from 0 to 1 in steps of 0.2). Finally, the individual protein score cut-offs derived for the in silico bioactivity application used in this work are released, as well as the reproducible and transferable KNIME workflows used to carry out the analysis.

Global Mapping of Traditional Chinese Medicine into Bioactivity Space and Pathways Annotation Improves Mechanistic Understanding and Discovers Relationships between Therapeutic Action (Sub)classes

Traditional Chinese medicine (TCM) still needs more scientific rationale to be proven for it to be accepted further in the West. We are now in the position to propose computational hypotheses for the mode-of-actions (MOAs) of 45 TCM therapeutic action (sub)classes from in silico target prediction algorithms, whose target was later annotated with Kyoto Encyclopedia of Genes and Genomes pathway, and to discover the relationship between them by generating a hierarchical clustering. The results of 10,749 TCM compounds showed 183 enriched targets and 99 enriched pathways from Estimation Score ≤ 0 and ≥ 5% of compounds/targets in a (sub)class. The MOA of a (sub)class was established from supporting literature. Overall, the most frequent top three enriched targets/pathways were immune-related targets such as tyrosine-protein phosphatase nonreceptor type 2 (PTPN2) and digestive system such as mineral absorption. We found two major protein families, G-protein coupled receptor (GPCR), and protein kinase family contributed to the diversity of the bioactivity space, while digestive system was consistently annotated pathway motif, which agreed with the important treatment principle of TCM, “the foundation of acquired constitution” that includes spleen and stomach. In short, the TCM (sub)classes, in many cases share similar targets/pathways despite having different indications.

Jaqpot Quattro: A Novel Computational Web Platform for Modeling and Analysis in Nanoinformatics

Engineered nanomaterials (ENMs) are increasingly infiltrating our lives as a result of their applications across multiple fields. However, ENM formulations may result in the modulation of pathways and mechanisms of toxic action that endanger human health and the environment. Alternative testing methods such as in silico approaches are becoming increasingly popular for assessing the safety of ENMs, as they are cost- and time-effective. Additionally, computational approaches support the industrial safer-by-design challenge and the REACH legislation objective of reducing animal testing. Because of the novelty of the field, there is also an evident need for harmonization in terms of databases, ontology, and modeling infrastructures. To this end, we present Jaqpot Quattro, a comprehensive open-source web application for ENM modeling with emphasis on predicting adverse effects of ENMs. We describe the system architecture and outline the functionalities, which include nanoQSAR modeling, validation services, read-across predictions, optimal experimental design, and interlaboratory testing.

Decision Trees for Continuous Data and Conditional Mutual Information as a Criterion for Splitting Instances

Decision trees are renowned in the computational chemistry and machine learning communities for their interpretability. Their capacity and usage are somewhat limited by the fact that they normally work on categorical data. Improvements to known decision tree algorithms are usually carried out by increasing and tweaking parameters, as well as the post-processing of the class assignment. In this work we attempted to tackle both these issues. Firstly, conditional mutual information was used as the criterion for selecting the attribute on which to split instances. The algorithm performance was compared with the results of C4.5 (WEKAs J48) using default parameters and no restrictions. Two datasets were used for this purpose, DrugBank compounds for HRH1 binding prediction and Traditional Chinese Medicine formulation predicted bioactivities for therapeutic class annotation. Secondly, an automated binning method for continuous data was evaluated, namely Scotts normal reference rule, in order to allow any decision tree to easily handle continuous data. This was applied to all approved drugs in DrugBank for predicting the RDKit SLogP property, using the remaining RDKit physicochemical attributes as input.

Comparative mode-of-action analysis following manual and automated phenotype detection in Xenopus laevis

Given the increasing utilization of phenotypic screens in drug discovery also the subsequent mechanism-of-action analysis gains increased attention. Such analyses frequently use in silico methods, which have become significantly more popular in recent years. However, identifying phenotype-specific mechanisms of action depends heavily on suitable phenotype identification in the first place, many of which rely on human input and are therefore inconsistent. In this work, we aimed at analysing the impact that human phenotype classification has on subsequent in silico mechanism-of-action analysis. To this end, an image analysis application was implemented for the rapid identification of seven high-level phenotypes in Xenopus laevis tadpoles treated with compounds from the National Cancer Institute Diversity Set II. It was found that manual and automated phenotype classifications were in agreement with some of the phenotypes (e.g. 73.9% agreement observed for general morphology abnormality), while this was not the case in others (e.g. melanophore migration with 37.6% agreement between both annotations). Based on both annotations, protein targets of active compounds were predicted in silico, and decision trees were generated to understand mechanisms-of-action behind every phenotype while also taking polypharmacology (combinations of targets) into account. It was found that the automated phenotype categorisation greatly increased the accuracy of the results of the mechanism-of-action model, where it improved the classification accuracy by 9.4%, as well as reducing the tree size by eight nodes and the number of leaves and the depth by three levels. Overall we conclude that consistent phenotype annotations seem to be generally crucial for successful subsequent mechanism-of-action analysis, and this is what we have shown here in Xenopus laevis screens in combination with in silico mechanism-of-action analysis.

Using machine learning techniques for rationalising phenotypic readouts from a rat sleeping model

Understanding the mode of action of small molecules is critical for drug research, both with respect to efficacy and anticipated side effects. Given that many compounds act on multiple targets simultaneously, it appears that linking single targets to outcomes is no longer sufficient. Hence, in this work we explore machine learning methods for rationalising phenotypic readouts from a rat model for hypnotics based on a polypharmacology approach. We hypothesise that by combining target prediction and machine learning techniques we are able to derive information regarding the mode of action of small molecules. In particular, we applied this hypothesis on a subset of the Eli Lilly SCORE™ dataset. This comprised 845 data instances, each consisting of 7 phenotypic readouts which attribute towards a good sleeping pattern. We employed a target prediction tool[1] to anticipate bioactivities of ligands in combination with the CN2 and C4.5 machine learning algorithms to derive interpretable rule lists and classification trees for the observed phenotypes. A review of the known mechanisms of action for the largest categories of hypnotics suggests that our results are in most cases consistent with current literature on the mode of action of hypnotics. This suggests that our method can potentially yield significant information regarding the mode of action of hypnotics, and in addition novel targets that are not yet well-established in literature. As further applications of this work, we are currently preparing to apply our methodology to a subset of a Traditional Chinese Medicine dataset and a phenotypic screening dataset for Xenopus.

Chapter 11:Computational Modelling of Biological Responses to Engineered Nanomaterials

In this chapter, we provide an overview of recent advancements related to the safety assessment of engineered nanomaterials (ENMs) using alternatives to animal testing strategies. Advanced risk assessment computational procedures include new methods for characterizing and describing the complex structures of ENMs, development of computational models predicting adverse effects, extension of “read-across” approaches taking into account different aspects of ENM similarity, integration of various testing strategies using a “weight-of-evidence” approach, and using omics data and pathways analysis technologies to provide insights into ENM mechanisms that potentially could induce toxicity.