Activity Prediction Of Compounds Research Articles

Benchmarking compound activity prediction for real-world drug discovery applications

Identifying active compounds for target proteins is fundamental in early drug discovery. Recently, data-driven computational methods have demonstrated promising potential in predicting compound activities. However, there lacks a well-designed benchmark to comprehensively evaluate these methods from a practical perspective. To fill this gap, we propose a Compound Activity benchmark for Real-world Applications (CARA). Through carefully distinguishing assay types, designing train-test splitting schemes and selecting evaluation metrics, CARA can consider the biased distribution of current real-world compound activity data and avoid overestimation of model performances. We observed that although current models can make successful predictions for certain proportions of assays, their performances varied across different assays. In addition, evaluation of several few-shot training strategies demonstrated different performances related to task types. Overall, we provide a high-quality dataset for developing and evaluating compound activity prediction models, and the analyses in this work may inspire better applications of data-driven models in drug discovery.

Enhancing property and activity prediction and interpretation using multiple molecular graph representations with MMGX

Graph Neural Networks (GNNs) excel in compound property and activity prediction, but the choice of molecular graph representations significantly influences model learning and interpretation. While atom-level molecular graphs resemble natural topology, they overlook key substructures or functional groups and their interpretation partially aligns with chemical intuition. Recent research suggests alternative representations using reduced molecular graphs to integrate higher-level chemical information and leverages both representations for model. However, there is a lack of studies about applicability and impact of different molecular graphs on model learning and interpretation. Here, we introduce MMGX (Multiple Molecular Graph eXplainable discovery), investigating the effects of multiple molecular graphs, including Atom, Pharmacophore, JunctionTree, and FunctionalGroup, on model learning and interpretation with various perspectives. Our findings indicate that multiple graphs relatively improve model performance, but in varying degrees depending on datasets. Interpretation from multiple graphs in different views provides more comprehensive features and potential substructures consistent with background knowledge. These results help to understand model decisions and offer valuable insights for subsequent tasks. The concept of multiple molecular graph representations and diverse interpretation perspectives has broad applicability across tasks, architectures, and explanation techniques, enhancing model learning and interpretation for relevant applications in drug discovery.

Compound Activity Prediction with Dose-Dependent Transcriptomic Profiles and Deep Learning.

Predicting compound activity in assays is a long-standing challenge in drug discovery. Computational models based on compound-induced gene expression signatures from a single profiling assay have shown promise toward predicting compound activity in other, seemingly unrelated, assays. Applications of such models include predicting mechanisms-of-action (MoA) for phenotypic hits, identifying off-target activities, and identifying polypharmacologies. Here, we introduce transcriptomics-to-activity transformer (TAT) models that leverage gene expression profiles observed over compound treatment at multiple concentrations to predict the compound activity in other biochemical or cellular assays. We built TAT models based on gene expression data from a RASL-seq assay to predict the activity of 2692 compounds in 262 dose-response assays. We obtained useful models for 51% of the assays, as determined through a realistic held-out set. Prospectively, we experimentally validated the activity predictions of a TAT model in a malaria inhibition assay. With a 63% hit rate, TAT successfully identified several submicromolar malaria inhibitors. Our results thus demonstrate the potential of transcriptomic responses over compound concentration and the TAT modeling framework as a cost-efficient way to identify the bioactivities of promising compounds across many assays.

EFFECTS OF LIMA BEAN (Phaseolus lunatus) FLOUR ON COGNITIVE FUNCTION AND GROWTH RECOVERY MALNUTRITION RATS

Incorporating lima bean (Phaseolus lunatus) flour can replace and complement the content of proteins, macronutrients, and micronutrients, which may be related to curative effects for malnutrition. Therefore, an in vivo approach was undertaken to demonstrate the effect of lima bean flour on cognitive function and growth recovery in malnourished rats. Four groups of male Wistar rats were included in this study: The control group (N), MAL, MAL+25% LB group, and MAL+50% LB group. Morphometric analysis was used to assess growth recovery. Cognitive tests included Y-Maze, Novel Object Recognition Test (NORT), and Water Maze by Morris (MWM) test, and histopathological examination of the brain sections was conducted on the cerebral cortex and hippocampus by hematoxylin and eosin stains. The biological activity prediction of lima bean active compounds was analyzed by the PASS server using an in-silico study. This study found that lima bean flour increased body weight, length, tail increment, ear length, and leg length in malnourished rats at 50% lima bean concentration, similar to the control group. Lima bean flour showed neuroprotective capabilities by reversing the effects of malnutrition on the cerebral cortex and hippocampus by lowering the percentage of degenerated cells and increasing cognitive and behavioral. Eight bioactive compounds in lima bean extract have antioxidant, anti-inflammatory, and antibacterial properties. Lima bean flour, a source of complex nutrients, protects against malnutrition systemically in various organ systems, including the brain.

Meta-learning for transformer-based prediction of potent compounds

For many machine learning applications in drug discovery, only limited amounts of training data are available. This typically applies to compound design and activity prediction and often restricts machine learning, especially deep learning. For low-data applications, specialized learning strategies can be considered to limit required training data. Among these is meta-learning that attempts to enable learning in low-data regimes by combining outputs of different models and utilizing meta-data from these predictions. However, in drug discovery settings, meta-learning is still in its infancy. In this study, we have explored meta-learning for the prediction of potent compounds via generative design using transformer models. For different activity classes, meta-learning models were derived to predict highly potent compounds from weakly potent templates in the presence of varying amounts of fine-tuning data and compared to other transformers developed for this task. Meta-learning consistently led to statistically significant improvements in model performance, in particular, when fine-tuning data were limited. Moreover, meta-learning models generated target compounds with higher potency and larger potency differences between templates and targets than other transformers, indicating their potential for low-data compound design.

Explaining Multiclass Compound Activity Predictions Using Counterfactuals and Shapley Values.

Most machine learning (ML) models produce black box predictions that are difficult, if not impossible, to understand. In pharmaceutical research, black box predictions work against the acceptance of ML models for guiding experimental work. Hence, there is increasing interest in approaches for explainable ML, which is a part of explainable artificial intelligence (XAI), to better understand prediction outcomes. Herein, we have devised a test system for the rationalization of multiclass compound activity prediction models that combines two approaches from XAI for feature relevance or importance analysis, including counterfactuals (CFs) and Shapley additive explanations (SHAP). For compounds with different single- and dual-target activities, we identified small compound modifications that induce feature changes inverting class label predictions. In combination with feature mapping, CFs and SHAP value calculations provide chemically intuitive explanations for model decisions.

Machine Learning-Based Quantitative Structure-Activity Relationship and ADMET Prediction Models for ERα Activity of Anti-Breast Cancer Drug Candidates

Breast cancer is presently one of the most common malignancies worldwide, with a higher fatality rate. In this study, a quantitative structure-activity relationship (QSAR) model of compound biological activity and ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) properties prediction model were performed using estrogen receptor alpha (ERα) antagonist information collected from compound samples. We first utilized grey relation analysis (GRA) in conjunction with the random forest (RF) algorithm to identify the top 20 molecular descriptor variables that have the greatest influence on biological activity, and then we used Spearman correlation analysis to identify 16 independent variables. Second, a QSAR model of the compound were developed based on BP neural network (BPNN), genetic algorithm optimized BP neural network (GA-BPNN), and support vector regression (SVR). The BPNN, the SVR, and the logistic regression (LR) models were then used to identify and predict the ADMET properties of substances, with the prediction impacts of each model compared and assessed. The results reveal that a SVR model was used in QSAR quantitative prediction, and in the classification prediction of ADMET properties: the SVR model predicts the Caco-2 and hERG(human Ether-a-go-go Related Gene) properties, the LR model predicts the cytochrome P450 enzyme 3A4 subtype (CYP3A4) and Micronucleus (MN) properties, and the BPNN model predicts the Human Oral Bioavailability (HOB) properties. Finally, information entropy theory is used to validate the rationality of variable screening, and sensitivity analysis of the model demonstrates that the constructed model has high accuracy and stability, which can be used as a reference for screening probable active compounds and drug discovery.

VDDB: A comprehensive resource and machine learning tool for antiviral drug discovery

AbstractVirus infection is one of the major diseases that seriously threaten human health. To meet the growing demand for mining and sharing data resources related to antiviral drugs and to accelerate the design and discovery of new antiviral drugs, we presented an open‐access antiviral drug resource and machine learning platform (VDDB), which, to the best of our knowledge, is the first comprehensive dedicated resource for experimentally verified potential drugs/molecules based on manually curated data. Currently, VDDB highlights 848 clinical vaccines and 199 clinical antibodies, as well as over 710,000 small molecules targeting 39 medically important viruses including severe acute respiratory syndrome coronavirus 2. Furthermore, VDDB stores approximately three million records of pharmacological data for these collected potential antiviral drugs/molecules, involving 314 cell infection‐based phenotypic and 234 target‐based genotypic assays. Based on these annotated pharmacological data, VDDB allows users to browse, search, and download reliable information about these collects for various viruses of interest. In particular, VDDB also integrates 57 cell infection‐ and 117 target‐based associated high‐accuracy machine learning models to support various antivirals identification‐related tasks, such as compound activity prediction, virtual screening, drug repositioning, and target fishing. VDDB is freely accessible at https://vddb.idruglab.cn.

Minimal data requirements for accurate compound activity prediction using machine learning methods of different complexity

Data sparseness is expected to negatively impact machine learning (ML) and especially deep learning (DL). In the era of DL, much emphasis is put on learning from large datasets. In cheminformatics, compound classification represents a suitable task for exploring characteristics of ML/DL. Here, we determine minimal data requirements for learning using activity-based compound classification as an exemplary application. Systematic predictions using message-passing neural networks, random forests, and simple controls are carried out on more than 100 activity classes using learning sets of increasing size. Contrary to expectations, predictive models are already obtained on the basis of “ultra-small” training sets containing only a few active compounds. These findings are rationalized by explainable ML, which reveals features that determine predictions. Accurate predictions often only depend on a few molecular features that might be detected even in ultra-small training sets. The findings revise current assumptions about data requirements for effective ML/DL.

EdgeSHAPer: Bond-centric Shapley value-based explanation method for graph neural networks

SummaryGraph neural networks (GNNs) recursively propagate signals along the edges of an input graph, integrate node feature information with graph structure, and learn object representations. Like other deep neural network models, GNNs have notorious black box character. For GNNs, only few approaches are available to rationalize model decisions. We introduce EdgeSHAPer, a generally applicable method for explaining GNN-based models. The approach is devised to assess edge importance for predictions. Therefore, EdgeSHAPer makes use of the Shapley value concept from game theory. For proof-of-concept, EdgeSHAPer is applied to compound activity prediction, a central task in drug discovery. EdgeSHAPer’s edge centricity is relevant for molecular graphs where edges represent chemical bonds. Combined with feature mapping, EdgeSHAPer produces intuitive explanations for compound activity predictions. Compared to a popular node-centric and another edge-centric GNN explanation method, EdgeSHAPer reveals higher resolution in differentiating features determining predictions and identifies minimal pertinent positive feature sets.

MLP-Based Regression Prediction Model For Compound Bioactivity.

The development of breast cancer is closely linked to the estrogen receptor ERα, which is also considered to be an important target for the treatment of breast cancer. Therefore, compounds that can antagonize ERα activity may be drug candidates for the treatment of breast cancer. In drug development, to save manpower and resources, potential active compounds are often screened by establishing compound activity prediction model. For the 1974 compounds collected, the top 20 molecular descriptors that significantly affected the biological activity were screened using LASSO regression models combined with 10-fold cross-validation method. Further, a regression prediction model based on the MLP fully connected neural network was constructed to predict the bioactivity values of 50 new compounds. To measure the validity of the model, the model loss term was specified as the mean squared error (MSE). The results showed that the MLP-based regression prediction model had a loss value of 0.0146 on the validation set. This model is therefore well trained and the prediction strategy used is valid. The methods developed by this paper may provide a reference for the development of anti-breast cancer drugs.

Review: Bioavailability and Activity Prediction of Bioactive Compounds of Red Fruit (Pandanus conoideus Lam.) and Pandan Grape (Sararanga sinousa Hemsley) by in silico Method

Red Fruit (RF) and Pandan grape (PG) contained bioactive compounds that good for health. The structure influenced the availability and activity, including the interaction with another in the body. It needs time and financial consumption, whereas bioinformation technology offers the solution. The research aims to predict the availability and activity of RF and PG compounds by in silico method based on the SAR. Fourteen RF and PF phenolics were evaluated for availability, antioxidant, and antiglycation activity during binding to HSA. The flavonoids showed Pa>0,7 for antioxidant in general, free radical scavenging, and lipid peroxidase inhibitor activity. The availability was shown by the ligand’s capability to bind to HSA. The order of affinity energy from the largest to the smallest was RF flavonoid > PG anthocyanin > GP phenolic acids. Taxifolin and Quercetin still had antioxidant activity during binding because there are free hydroxyl groups. PG chlorogenic acid and RF flavonoids play as an anti-diabetic through antioxidant mechanism and HSA glycation disturbance. Based on this review, in silico method is effective as an analysis tool for the activity and mechanism prediction of food bioactive compounds.

Study on Quantitative Prediction of Anticancer Compound Activity Based on Neural Network Model

Breast cancer poses a serious threat to global women's health. Around this problem, a quantitative prediction model of biological activity based on neural network and a quantitative prediction model of multiple linear regression are established in this paper. 20 kinds of molecular descriptor data with the most significant effect on biological activity of 1974 compounds were selected as input layer and 1 biological activity data as output layer, and a quantitative prediction model of biological activity based on neural network was established. Based on the quantitative prediction model based on parameter setting, the single hidden layer BP neural network structure is used to test the model, and the fitting error R values of the training set and test set of the model are close to 1, and the trend in the scatter diagram is obvious, which shows that the prediction value of the model is more accurate.

Efficient Compound Activity Prediction Model Based on Bayesian Regularization Method for Candidate Drugs

In breast cancer diagnoses, antagonist compounds are used to resist <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula> Subtype of Estrogen Receptor (ER <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\alpha$ </tex-math></inline-formula> ) bioactivity. However, those compounds are difficult to be obtained in the process of drug screening. In this paper, an efficient bioactivity prediction model is proposed with consideration of Absorption, Distribution, Metabolism, Excretion and Toxicity (ADMET) properties as an auxiliary prediction to achieve the minimization of the damage caused to human body whereas maximizing the efficacy of estrogen activity inhibitors. In the proposed prediction model, Pearson correlation analysis and Bayesian regularization algorithm under neural network are specifically used to train and analyze the data of bioactivity. To calculate the loss of neural network, cross-entropy and supervise learning are applied. The results show that the accuracy of our proposed compound prediction model can reach up to 92.2%, 94.3%, and 90.2% for Caco2, CYP3A4, and hERG, respectively. Moreover, the Area Under the Curve (AUC) value of our proposed model is 0.9752, which is 13-14% higher than that of SVM method. In addition, the Matthews Correlation Coefficient (MCC) can reach up to 0.93 ± 0.03, 37-40% higher than that of SVM method. Due to correlation analysis and valid data screening, our model can shorten the prediction time by 15%. Due to its advantages in accuracy and speed of prediction, this research can contribute to the upgrading of diagnosis and treatment of breast cancer.

In Silico Evaluation of Antidiabetic Activity and ADMET Prediction of Compounds from Musa acuminata Colla Peel

Diabetes is a severe chronic disease that affects 422 million adults worldwide in 2014. It was the fifth leading cause of mortality in the Philippines in 2019. The primary cause of death of diabetic patients is due to cardiovascular disease. α-Glucosidase inhibitors (AGIs) are known for their cardiovascular benefits because they prevent the postprandial glucose level from increasing, which plays a significant role in the development of cardiovascular diseases. The synthesis of AGIs is complex and requires a lot of steps. Thus, there is a need to explore and discover AGIs especially from plants, which are known sources of bioactive compounds. Drug discovery entails a complex, costly, time-consuming, and risky process. Computer-aided drug discovery/ design (CADD) methods such as molecular docking and ADMET (absorption, distribution, metabolism, excretion, and toxicity) prediction have been developed to identify the promising compounds that will be tested in in vitro and in vivo experiments. This shortens the research process and helps reduce the expense and risk of failure for drug discovery. In this study, molecular docking was conducted to predict the α-glucosidase inhibitory activity of compounds from Musa acuminata Colla peel against human intestinal α-glucosidase. Out of 87 compounds, only 11 compounds were found to have better or comparable binding affinity with the standard, acarbose (–8.8 kcal/mol) – namely, sesamin (–9.8 kcal/mol), asarinin (–9.7 kcal/mol), quercetin- 7-rutinoside (–9.4 kcal/mol), kaempferol-3-rutinoside (–9.4 kcal/mol), (–)-epicatechin (–9.0 kcal/ mol), (+)-catechin (–8.9 kcal/mol), myricetin-3-rutinoside (–8.9 kcal/mol), quercetin (–8.9 kcal/ mol), kaempferol-3-rhamnoside-7-glucoside (–8.9 kcal/mol), stigmasterol (–8.9 kcal/mol), and β-sitosterol (-8.8 Kcal/mol). The prediction of ADMET properties and drug-likeness revealed how the best binding compounds may behave inside the body. Some of the compounds were found to be safe and have good absorption, distribution, metabolism, and excretion properties. They showed promising potentials that may lead to their development as drugs.

End-to-end learning for compound activity prediction based on binding pocket information

BackgroundRecently, machine learning-based ligand activity prediction methods have been greatly improved. However, if known active compounds of a target protein are unavailable, the machine learning-based method cannot be applied. In such cases, docking simulation is generally applied because it only requires a tertiary structure of the target protein. However, the conformation search and the evaluation of binding energy of docking simulation are computationally heavy and thus docking simulation needs huge computational resources. Thus, if we can apply a machine learning-based activity prediction method for a novel target protein, such methods would be highly useful. Recently, Tsubaki et al. proposed an end-to-end learning method to predict the activity of compounds for novel target proteins. However, the prediction accuracy of the method was still insufficient because it only used amino acid sequence information of a protein as the input.ResultsIn this research, we proposed an end-to-end learning-based compound activity prediction using structure information of a binding pocket of a target protein. The proposed method learns the important features by end-to-end learning using a graph neural network both for a compound structure and a protein binding pocket structure. As a result of the evaluation experiments, the proposed method has shown higher accuracy than an existing method using amino acid sequence information.ConclusionsThe proposed method achieved equivalent accuracy to docking simulation using AutoDock Vina with much shorter computing time. This indicated that a machine learning-based approach would be promising even for novel target proteins in activity prediction.

Mutual Support of Ligand- and Structure-Based Approaches-To What Extent We Can Optimize the Power of Predictive Model? Case Study of Opioid Receptors.

The process of modern drug design would not exist in the current form without computational methods. They are part of every stage of the drug design pipeline, supporting the search and optimization of new bioactive substances. Nevertheless, despite the great help that is offered by in silico strategies, the power of computational methods strongly depends on the input data supplied at the stage of the predictive model construction. The studies on the efficiency of the computational protocols most often focus on global efficiency. They use general parameters that refer to the whole dataset, such as accuracy, precision, mean squared error, etc. In the study, we examined machine learning predictions obtained for opioid receptors (mu, kappa, delta) and focused on cases for which the predictions were the most accurate and the least accurate. Moreover, by using docking, we tried to explain prediction errors. We attempted to develop a rule of thumb, which can help in the prediction of compound activity towards opioid receptors via docking, especially those that have been incorrectly predicted by machine learning. We found out that although the combination of ligand- and structure-based path can be beneficial for the prediction accuracy, there still remain cases that cannot be reliably predicted by any available modeling method. In addition to challenging ligand- and structure-based predictions, we also examined the role of the application of machine-learning methods in comparison to simple statistical methods for both standard ligand-based representations (molecular fingerprints) and interaction fingerprints. All approaches were confronted in both classification (where compounds were assigned to the group of active and inactive group constructed on the basis of Ki values) and regression (where exact Ki value was predicted) experiments.

Antiviral Activity and Toxicity Prediction of Compounds Contained in Figs (Ficus carica L.) by In Silico Method

Viral infection is a global health problem that can cause endemic to pandemic. Compounds delivered from plants has been developed as an alternative antiviral agent. One of the plants that can be used as antiviral therapy is Ficus carica L. (figs). The aim of this research is to predict the inhibitory activity and toxicity of compounds contained in figs as an antiviral for HIV-1 using in silico method. Compounds were docked to the HIV-1 Reverse Transcriptase protein (PDB ID: 3LAL). Threedimentional structures were modeled using GaussView and optimized using Gaussian 09W. Optimized compounds were docked to the target protein using AutoDock Tools and the interaction to protein binding side were analyzed in comparison with the standard compounds. The standard compound used for the analysis nevirapine, efavirenz, and doravirine. The compound toxicity was analyzed using ECOSAR and Toxtree. Based on the results, the compounds that has similar interaction to the standard compounds were campesterol which has 4 similar hydrophobic interactions. Based on the classification of Cramer Rules for toxicity test, campesterol are classified in class 3 (high toxicity) and according to the Benigni/Bossa Rulebase classification, campesterol are negative for genotoxic and nongenotoxic carcinogenicity.Keywords: Antivirus, HIV, figs, molecular docking, toxicity

Activity Prediction of Bioactive Compounds Contained in Etlingera elatior Against the SARS-CoV-2 Main Protease: An In Silico Approach

The COVID-19 pandemic has become a serious problem today, with its prevalence increasing every day. The SARS-CoV-2 main protease (MPro) is a promising therapeutic target to inhibit replicating and spreading the virus that causes COVID-19. The compounds contained in the Etlingera elatior plant has the potential. This study aimed to examine the compounds' activity in E. elatior against SARS-CoV-2 MPro using in silico methods. A total of seven compounds contained in E. elatior were obtained from the Knapsack database. The compounds were then docked into the SARS-CoV-2 MPro receptor's active site with the PDB ID 6LU7. Afterward, the biological activities were predicted by the PASS prediction webserver. The molecular docking results showed that ergosterol peroxide and sitostenone had the best binding energy with -10.40 kcal/mol and -9.17 kcal/mol, respectively. The in silico PASS prediction showed it has potential as antiviral therapy. It concluded ergosterol peroxide and sitostenone has the potential as SARS-CoV-2 MPro inhibitor candidate.

Data Integration Using Advances in Machine Learning in Drug Discovery and Molecular Biology.

While the term artificial intelligence and the concept of deep learning are not new, recent advances in high-performance computing, the availability of large annotated data sets required for training, and novel frameworks for implementing deep neural networks have led to an unprecedented acceleration of the field of molecular (network) biology and pharmacogenomics. The need to align biological data to innovative machine learning has stimulated developments in both data integration (fusion) and knowledge representation, in the form of heterogeneous, multiplex, and biological networks or graphs. In this chapter we briefly introduce several popular neural network architectures used in deep learning, namely, the fully connected deep neural network, recurrent neural network, convolutional neural network, and the autoencoder. Deep learning predictors, classifiers, and generators utilized in modern feature extraction may well assist interpretability and thus imbue AI tools with increased explication, potentially adding insights and advancements in novel chemistry and biology discovery.The capability of learning representations from structures directly without using any predefined structure descriptor is an important feature distinguishing deep learning from other machine learning methods and makes the traditional feature selection and reduction procedures unnecessary. In this chapter we briefly show how these technologies are applied for data integration (fusion) and analysis in drug discovery research covering these areas: (1) application of convolutional neural networks to predict ligand-protein interactions; (2) application of deep learning in compound property and activity prediction; (3) de novo design through deep learning. We also: (1) discuss some aspects of future development of deep learning in drug discovery/chemistry; (2) provide references to published information; (3) provide recently advocated recommendations on using artificial intelligence and deep learning in -omics research and drug discovery.