MEDICASCY: A Machine Learning Approach for Predicting Small-Molecule Drug Side Effects, Indications, Efficacy, and Modes of Action

On the additive artificial intelligence-based discovery of nanoparticle neurodegenerative disease drug delivery systems

Neurodegenerative diseases are characterized by slowly progressing neuronal cell death. Conventional drug treatment strategies often fail because of poor solubility, low bioavailability, and the inability of the drugs to effectively cross the blood-brain barrier. Therefore, the development of new neurodegenerative disease drugs (NDDs) requires immediate attention. Nanoparticle (NP) systems are of increasing interest for transporting NDDs to the central nervous system. However, discovering effective nanoparticle neuronal disease drug delivery systems (N2D3Ss) is challenging because of the vast number of combinations of NP and NDD compounds, as well as the various assays involved. Artificial intelligence/machine learning (AI/ML) algorithms have the potential to accelerate this process by predicting the most promising NDD and NP candidates for assaying. Nevertheless, the relatively limited amount of reported data on N2D3S activity compared to assayed NDDs makes AI/ML analysis challenging. In this work, the IFPTML technique, which combines information fusion (IF), perturbation theory (PT), and machine learning (ML), was employed to address this challenge. Initially, we conducted the fusion into a unified dataset comprising 4403 NDD assays from ChEMBL and 260 NP cytotoxicity assays from journal articles. Through a resampling process, three new working datasets were generated, each containing 500,000 cases. We utilized linear discriminant analysis (LDA) along with artificial neural network (ANN) algorithms, such as multilayer perceptron (MLP) and deep learning networks (DLN), to construct linear and non-linear IFPTML models. The IFPTML-LDA models exhibited sensitivity (Sn) and specificity (Sp) values in the range of 70% to 73% (>375,000 training cases) and 70% to 80% (>125,000 validation cases), respectively. In contrast, the IFPTML-MLP and IFPTML-DLN achieved Sn and Sp values in the range of 85% to 86% for both training and validation series. Additionally, IFPTML-ANN models showed an area under the receiver operating curve (AUROC) of approximately 0.93 to 0.95. These results indicate that the IFPTML models could serve as valuable tools in the design of drug delivery systems for neurosciences.

Machine learning-based integration of network features and chemical structure of compounds for SARS-CoV-2 drug effect analysis

High drug development costs and the limited number of new annual drug approvals increase the need for innovative approaches for drug effect prediction. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the cause of coronavirus disease 2019 (COVID-19), led to a global pandemic with high morbidity and mortality. Although effective preventive measures exist, there are few effective treatments for hospitalized patients with SARS-CoV-2 infection. Drug repurposing and drug effect prediction are promising strategies that could shorten development time and reduce costs compared with de novo drug discovery. In this work, we present a machine learning framework to integrate a variety of target network features and physicochemical properties of compounds, and analyze their influence on the therapeutic effects for SARS-CoV-2 infection and on host cell cytotoxic effects. Random forest models trained on compounds with known experimental effects on SARS-CoV-2 infection and subsequent feature importance analysis based on Shapley values provided insights into the determinants of drug efficacy and cytotoxicity, which can be incorporated into novel drug discovery approaches. Given the complexity of molecular mechanisms of drug action and limited sample sizes, our models achieve a reasonable mean area under the receiver operating characteristic curve (ROC-AUC) of 0.73 on an unseen validation set. To our knowledge, this is the first work to incorporate a combination of network and physicochemical features of compounds into a machine learning model to predict drug effects on SARS-CoV-2 infection. Our systems pharmacology-based machine learning framework can be used to classify other existing drugs for SARS-CoV-2 infection and can easily be adapted to drug effect prediction for future viral outbreaks.

Non-Negative matrix factorization combined with kernel regression for the prediction of adverse drug reaction profiles

Motivation

Post-market unexpected Adverse Drug Reactions (ADRs) are associated with significant costs, in both financial burden and human health. Due to the high cost and time required to run clinical trials, there is significant interest in accurate computational methods that can aid in the prediction of ADRs for new drugs. As a machine learning task, ADR prediction is made more challenging due to a high degree of class imbalance and existing methods do not successfully balance the requirement to detect the minority cases (true positives for ADR), as measured by the Area Under the Precision-Recall (AUPR) curve with the ability to separate true positives from true negatives [as measured by the Area Under the Receiver Operating Characteristic (AUROC) curve]. Surprisingly, the performance of most existing methods is worse than a naïve method that attributes ADRs to drugs according to the frequency with which the ADR has been observed over all other drugs. The existing advanced methods applied do not lead to substantial gains in predictive performance.

Results

We designed a rigorous evaluation to provide an unbiased estimate of the performance of ADR prediction methods: Nested Cross-Validation and a hold-out set were adopted. Among the existing methods, Kernel Regression (KR) performed best in AUPR but had a disadvantage in AUROC, relative to other methods, including the naïve method. We proposed a novel method that combines non-negative matrix factorization with kernel regression, called VKR. This novel approach matched or exceeded the performance of existing methods, overcoming the weakness of the existing methods.

Availability

Code and data are available on https://github.com/YezhaoZhong/VKR.

TTD: Therapeutic Target Database describing target druggability information

Target discovery is one of the essential steps in modern drug development, and the identification of promising targets is fundamental for developing first-in-class drug. A variety of methods have emerged for target assessment based on druggability analysis, which refers to the likelihood of a target being effectively modulated by drug-like agents. In the therapeutic target database (TTD), nine categories of established druggability characteristics were thus collected for 426 successful, 1014 clinical trial, 212 preclinical/patented, and 1479 literature-reported targets via systematic review. These characteristic categories were classified into three distinct perspectives: molecular interaction/regulation, human system profile and cell-based expression variation. With the rapid progression of technology and concerted effort in drug discovery, TTD and other databases were highly expected to facilitate the explorations of druggability characteristics for the discovery and validation of innovative drug target. TTD is now freely accessible at: https://idrblab.org/ttd/.

Machine Learning Methods for Small Data Challenges in Molecular Science

Small data are often used in scientific and engineering research due to the presence of various constraints, such as time, cost, ethics, privacy, security, and technical limitations in data acquisition. However, big data have been the focus for the past decade, small data and their challenges have received little attention, even though they are technically more severe in machine learning (ML) and deep learning (DL) studies. Overall, the small data challenge is often compounded by issues, such as data diversity, imputation, noise, imbalance, and high-dimensionality. Fortunately, the current big data era is characterized by technological breakthroughs in ML, DL, and artificial intelligence (AI), which enable data-driven scientific discovery, and many advanced ML and DL technologies developed for big data have inadvertently provided solutions for small data problems. As a result, significant progress has been made in ML and DL for small data challenges in the past decade. In this review, we summarize and analyze several emerging potential solutions to small data challenges in molecular science, including chemical and biological sciences. We review both basic machine learning algorithms, such as linear regression, logistic regression (LR), k-nearest neighbor (KNN), support vector machine (SVM), kernel learning (KL), random forest (RF), and gradient boosting trees (GBT), and more advanced techniques, including artificial neural network (ANN), convolutional neural network (CNN), U-Net, graph neural network (GNN), Generative Adversarial Network (GAN), long short-term memory (LSTM), autoencoder, transformer, transfer learning, active learning, graph-based semi-supervised learning, combining deep learning with traditional machine learning, and physical model-based data augmentation. We also briefly discuss the latest advances in these methods. Finally, we conclude the survey with a discussion of promising trends in small data challenges in molecular science.

An extensive survey on the use of supervised machine learning techniques in the past two decades for prediction of drug side effects

Approved drugs for sale must be effective and safe, implying that the drug's advantages outweigh its known harmful side effects. Side effects (SE) of drugs are one of the common reasons for drug failure that may halt the whole drug discovery pipeline. The side effects might vary from minor concerns like a runny nose to potentially life-threatening issues like liver damage, heart attack, and death. Therefore, predicting the side effects of the drug is vital in drug development, discovery, and design. Supervised machine learning-based side effects prediction task has recently received much attention since it reduces time, chemical waste, design complexity, risk of failure, and cost. The advancement of supervised learning approaches for predicting side effects have emerged as essential computational tools. Supervised machine learning technique provides early information on drug side effects to develop an effective drug based on drug properties. Still, there are several challenges to predicting drug side effects. Thus, a near-exhaustive survey is carried out in this paper on the use of supervised machine learning approaches employed in drug side effects prediction tasks in the past two decades. In addition, this paper also summarized the drug descriptor required for the side effects prediction task, commonly utilized drug properties sources, computational models, and their performances. Finally, the research gap, open problems, and challenges for the further supervised learning-based side effects prediction task have been discussed.

Multimodal representation learning for predicting molecule-disease relations

MOTIVATION

Predicting molecule-disease indications and side effects is important for drug development and pharmacovigilance. Comprehensively mining molecule-molecule, molecule-disease and disease-disease semantic dependencies can potentially improve prediction performance.

METHODS

We introduce a Multi-Modal REpresentation Mapping Approach to Predicting molecular-disease relations (M2REMAP) by incorporating clinical semantics learned from electronic health records (EHR) of 12.6 million patients. Specifically, M2REMAP first learns a multimodal molecule representation that synthesizes chemical property and clinical semantic information by mapping molecule chemicals via a deep neural network onto the clinical semantic embedding space shared by drugs, diseases and other common clinical concepts. To infer molecule-disease relations, M2REMAP combines multimodal molecule representation and disease semantic embedding to jointly infer indications and side effects.

RESULTS

We extensively evaluate M2REMAP on molecule indications, side effects and interactions. Results show that incorporating EHR embeddings improves performance significantly, for example, attaining an improvement over the baseline models by 23.6% in PRC-AUC on indications and 23.9% on side effects. Further, M2REMAP overcomes the limitation of existing methods and effectively predicts drugs for novel diseases and emerging pathogens.

AVAILABILITY AND IMPLEMENTATION

The code is available at https://github.com/celehs/M2REMAP, and prediction results are provided at https://shiny.parse-health.org/drugs-diseases-dev/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

DeepSide: A Deep Learning Approach for Drug Side Effect Prediction

Drug failures due to unforeseen adverse effects at clinical trials pose health risks for the participants and lead to substantial financial losses. Side effect prediction algorithms have the potential to guide the drug design process. LINCS L1000 dataset provides a vast resource of cell line gene expression data perturbed by different drugs and creates a knowledge base for context specific features. The state-of-the-art approach that aims at using context specific information relies on only the high-quality experiments in LINCS L1000 and discards a large portion of the experiments. In this study, our goal is to boost the prediction performance by utilizing this data to its full extent. We experiment with 5 deep learning architectures. We find that a multi-modal architecture produces the best predictive performance among multi-layer perceptron-based architectures when drug chemical structure (CS), and the full set of drug perturbed gene expression profiles (GEX) are used as modalities. Overall, we observe that the CS is more informative than the GEX. A convolutional neural network-based model that uses only SMILES string representation of the drugs achieves the best results and provides 13.0% macro-AUC and 3.1% micro-AUC improvements over the state-of-the-art. We also show that the model is able to predict side effect-drug pairs that are reported in the literature but was missing in the ground truth side effect dataset. DeepSide is available at http://github.com/OnurUner/DeepSide.

PHEVIR: an artificial intelligence algorithm that predicts the molecular role of pathogens in complex human diseases

Infectious diseases are known to cause a wide variety of post-infection complications. However, it's been challenging to identify which diseases are most associated with a given pathogen infection. Using the recently developed LeMeDISCO approach that predicts comorbid diseases associated with a given set of putative mode of action (MOA) proteins and pathogen-human protein interactomes, we developed PHEVIR, an algorithm which predicts the corresponding human disease comorbidities of 312 viruses and 57 bacteria. These predictions provide an understanding of the molecular bases of complications and means of identifying appropriate drug targets to treat them. As an illustration of its power, PHEVIR is applied to identify putative driver pathogens and corresponding human MOA proteins for Type 2 diabetes, atherosclerosis, Alzheimer's disease, and inflammatory bowel disease. Additionally, we explore the origins of the oncogenicity/oncolyticity of certain pathogens and the relationship between heart disease and influenza. The full PHEVIR database is available at https://sites.gatech.edu/cssb/phevir/ .

LeMeDISCO is a computational method for large-scale prediction & molecular interpretation of disease comorbidity

To understand the origin of disease comorbidity and to identify the essential proteins and pathways underlying comorbid diseases, we developed LeMeDISCO (Large-Scale Molecular Interpretation of Disease Comorbidity), an algorithm that predicts disease comorbidities from shared mode of action proteins predicted by the artificial intelligence-based MEDICASCY algorithm. LeMeDISCO was applied to predict the occurrence of comorbid diseases for 3608 distinct diseases. Benchmarking shows that LeMeDISCO has much better comorbidity recall than the two molecular methods XD-score (44.5% vs. 6.4%) and the S_AB score (68.6% vs. 8.0%). Its performance is somewhat comparable to the phenotype method-based Symptom Similarity Score, 63.7% vs. 100%, but LeMeDISCO works for far more cases and its large comorbidity recall is attributed to shared proteins that can help provide an understanding of the molecular mechanism(s) underlying disease comorbidity. The LeMeDISCO web server is available for academic users at: http://sites.gatech.edu/cssb/LeMeDISCO .

The promising role of new molecular biomarkers in prostate cancer: from coding and non-coding genes to artificial intelligence approaches

Background

Risk stratification or progression in prostate cancer is performed with the support of clinical-pathological data such as the sum of the Gleason score and serum levels PSA. For several decades, methods aimed at the early detection of prostate cancer have included the determination of PSA serum levels. The aim of this systematic review is to provide an overview about recent advances in the discovery of new molecular biomarkers through transcriptomics, genomics and artificial intelligence that are expected to improve clinical management of the prostate cancer patient.

Methods

An exhaustive search was conducted by Pubmed, Google Scholar and Connected Papers using keywords relating to the genetics, genomics and artificial intelligence in prostate cancer, it includes “biomarkers”, “non-coding RNAs”, “lncRNAs”, “microRNAs”, “repetitive sequence”, “prognosis”, “prediction”, “whole-genome sequencing”, “RNA-Seq”, “transcriptome”, “machine learning”, and “deep learning”.

Results

New advances, including the search for changes in novel biomarkers such as mRNAs, microRNAs, lncRNAs, and repetitive sequences, are expected to contribute to an earlier and accurate diagnosis for each patient in the context of precision medicine, thus improving the prognosis and quality of life of patients. We analyze several aspects that are relevant for prostate cancer including its new molecular markers associated with diagnosis, prognosis, and prediction to therapy and how bioinformatic approaches such as machine learning and deep learning can contribute to clinic. Furthermore, we also include current techniques that will allow an earlier diagnosis, such as Spatial Transcriptomics, Exome Sequencing, and Whole-Genome Sequencing.

Conclusion

Transcriptomic and genomic analysis have contributed to generate knowledge in the field of prostate carcinogenesis, new information about coding and non-coding genes as biomarkers has emerged. Synergies created by the implementation of artificial intelligence to analyze and understand sequencing data have allowed the development of clinical strategies that facilitate decision-making and improve personalized management in prostate cancer.

Machine learning predicts the effect of food on orally administered medicines

Food-mediated changes to drug absorption, termed the food effect, are hard to predict and can have significant implications for the safety and efficacy of oral drug products in patients. Mimicking the prandial states of the human gastrointestinal tract in preclinical studies is challenging, poorly predictive and can produce difficult to interpret datasets. Machine learning (ML) has emerged from the computer science field and shows promise in interpreting complex datasets present in the pharmaceutical field. A ML-based approach aimed to predict the food effect based on an extensive dataset of over 311 drugs with more than 20 drug physicochemical properties, referred to as features. Machine learning techniques were tested; including logistic regression, support vector machine, k-Nearest neighbours and random forest. First a standard ML pipeline using a 80:20 split for training and testing was tried to predict no food effect, negative food effect and positive food effect, however this lead to specificities of less than 40%. To overcome this, a strategic ML pipeline was devised and three tasks were developed. Random forest achieved the strongest performance overall. High accuracies and sensitivities of 70%, 80% and 70% and specificities of 71%, 76% and 71% were achieved for classifying; (i) no food effect vs food effect, (ii) negative food vs positive food effect and (iii) no food effect vs negative food effect vs positive food effect, respectively. Feature importance using random forest ranked the features by importance for building the predictive tasks. The calculated dose number was the most important feature. Here, ML has provided an effective screening tool for predicting the food effect, with the potential to select lead compounds with no food effect, reduce the number of animal studies, and accelerate oral drug development studies.

Towards machine learning discovery of dual antibacterial drug-nanoparticle systems

Artificial Intelligence/Machine Learning (AI/ML) algorithms may speed up the design of DADNP systems formed by Antibacterial Drugs (AD) and Nanoparticles (NP). In this work, we used IFPTML = Information Fusion (IF) + Perturbation-Theory (PT) + Machine Learning (ML) algorithm for the first time to study of a large dataset of putative DADNP systems composed by >165 000 ChEMBL AD assays and 300 NP assays vs. multiple bacteria species. We trained alternative models with Linear Discriminant Analysis (LDA), Artificial Neural Networks (ANN), Bayesian Networks (BNN), K-Nearest Neighbour (KNN) and other algorithms. IFPTML-LDA model was simpler with values of Sp ≈ 90% and Sn ≈ 74% in both training (>124 K cases) and validation (>41 K cases) series. IFPTML-ANN and KNN models are notably more complicated even when they are more balanced Sn ≈ Sp ≈ 88.5%-99.0% and AUROC ≈ 0.94-0.99 in both series. We also carried out a simulation (>1900 calculations) of the expected behavior for putative DADNPs in 72 different biological assays. The putative DADNPs studied are formed by 27 different drugs with multiple classes of NP and types of coats. In addition, we tested the validity of our additive model with 80 DADNP complexes experimentally synthetized and biologically tested (reported in >45 papers). All these DADNPs show values of MIC < 50 μg mL^-1 (cutoff used) better that MIC of AD and NP alone (synergistic or additive effect). The assays involve DADNP complexes with 10 types of NP, 6 coating materials, NP size range 5-100 nm vs. 15 different antibiotics, and 12 bacteria species. The IFPTML-LDA model classified correctly 100% (80 out of 80) DADNP complexes as biologically active. IFPMTL additive strategy may become a useful tool to assist the design of DADNP systems for antibacterial therapy taking into consideration only information about AD and NP components by separate.

AlphaFold 2: Why It Works and Its Implications for Understanding the Relationships of Protein Sequence, Structure, and Function

AlphaFold 2 (AF2) was the star of CASP14, the last biannual structure prediction experiment. Using novel deep learning, AF2 predicted the structures of many difficult protein targets at or near experimental resolution. Here, we present our perspective of why AF2 works and show that it is a very sophisticated fold recognition algorithm that exploits the completeness of the library of single domain PDB structures. It has also learned local side chain packing rearrangements that enable it to refine proteins to high resolution. The benefits and limitations of its ability to predict the structures of many more proteins at or close to atomic detail are discussed.

Prediction of severe adverse events, modes of action and drug treatments for COVID-19's complications

Following SARS-CoV-2 infection, some COVID-19 patients experience severe host driven adverse events. To treat these complications, their underlying etiology and drug treatments must be identified. Thus, a novel AI methodology MOATAI-VIR, which predicts disease-protein-pathway relationships and repurposed FDA-approved drugs to treat COVID-19's clinical manifestations was developed. SARS-CoV-2 interacting human proteins and GWAS identified respiratory failure genes provide the input from which the mode-of-action (MOA) proteins/pathways of the resulting disease comorbidities are predicted. These comorbidities are then mapped to their clinical manifestations. To assess each manifestation's molecular basis, their prioritized shared proteins were subject to global pathway analysis. Next, the molecular features associated with hallmark COVID-19 phenotypes, e.g. unusual neurological symptoms, cytokine storms, and blood clots were explored. In practice, 24/26 of the major clinical manifestations are successfully predicted. Three major uncharacterized manifestation categories including neoplasms are also found. The prevalence of neoplasms suggests that SARS-CoV-2 might be an oncovirus due to shared molecular mechanisms between oncogenesis and viral replication. Then, repurposed FDA-approved drugs that might treat COVID-19's clinical manifestations are predicted by virtual ligand screening of the most frequent comorbid protein targets. These drugs might help treat both COVID-19's severe adverse events and lesser ones such as loss of taste/smell.

Transcriptional Regulation of Cancer Immune Checkpoints: Emerging Strategies for Immunotherapy

The study of immune evasion has gained a well-deserved eminence in cancer research by successfully developing a new class of therapeutics, immune checkpoint inhibitors, such as pembrolizumab and nivolumab, anti-PD-1 antibodies. By aiming at the immune checkpoint blockade (ICB), these new therapeutics have advanced cancer treatment with notable increases in overall survival and tumor remission. However, recent reports reveal that 40-60% of patients fail to benefit from ICB therapy due to acquired resistance or tumor relapse. This resistance may stem from increased expression of co-inhibitory immune checkpoints or alterations in the tumor microenvironment that promotes immune suppression. Because these mechanisms are poorly elucidated, the transcription factors that regulate immune checkpoints, known as "master regulators", have garnered interest. These include AP-1, IRF-1, MYC, and STAT3, which are known to regulate PD/PD-L1 and CTLA-4. Identifying these and other potential master regulators as putative therapeutic targets or biomarkers can be facilitated by mining cancer literature, public datasets, and cancer genomics resources. In this review, we describe recent advances in master regulator identification and characterization of the mechanisms underlying immune checkpoints regulation, and discuss how these master regulators of immune checkpoint molecular expression can be targeted as a form of auxiliary therapeutic strategy to complement traditional immunotherapy.