Predicting coated-nanoparticle drug release systems with perturbation-theory machine learning (PTML) models

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.

The drug loading capacity prediction and cytotoxicity analysis of metal-organic frameworks using stacking algorithms of machine learning

Metal-organic frameworks (MOFs) have shown excellent performance in the field of drug delivery. Despite the synthesis of a vast array of MOFs exceeding 100,000 varieties, certain formulations have exhibited suboptimal performance characteristics. Therefore, there is a pressing need to enhance their efficacy by identifying MOFs with superior drug loading capacities and minimal cytotoxicity, which can be achieved through machine learning (ML). In this study, a stacking regression model was developed to predict drug loading capacity and cytotoxicity of MOFs using datasets compiled from various literature sources. The model exhibited exceptional predictive capabilities, achieving R2 values of 0.907 for drug loading capacity and 0.856 for cytotoxicity. Furthermore, various model interpretation methods including partial dependence plots, individual conditional expectation, Shapley additive explanation, decision tree, random forest, CatBoost Regressor, and light gradient-boosting machine were employed for feature importance analysis. The results revealed that specific metal atoms such as Zn, Cr, Fe, Zr, and Cu significantly influenced the drug loading capacity and cytotoxicity of MOFs. Through model validation encompassing experimental validation and computational verification, the reliability of the model was thoroughly established. In general, it is a good practice to use ML methods for predicting drug loading capacity and cytotoxicity analysis of MOFs, guiding the development of future property prediction methods for MOFs.

MATEO: intermolecular α-amidoalkylation theoretical enantioselectivity optimization. Online tool for selection and design of chiral catalysts and products

The enantioselective Brønsted acid-catalyzed α-amidoalkylation reaction is a useful procedure is for the production of new drugs and natural products. In this context, Chiral Phosphoric Acid (CPA) catalysts are versatile catalysts for this type of reactions. The selection and design of new CPA catalysts for different enantioselective reactions has a dual interest because new CPA catalysts (tools) and chiral drugs or materials (products) can be obtained. However, this process is difficult and time consuming if approached from an experimental trial and error perspective. In this work, an Heuristic Perturbation-Theory and Machine Learning (HPTML) algorithm was used to seek a predictive model for CPA catalysts performance in terms of enantioselectivity in α-amidoalkylation reactions with R2 = 0.96 overall for training and validation series. It involved a Monte Carlo sampling of > 100,000 pairs of query and reference reactions. In addition, the computational and experimental investigation of a new set of intermolecular α-amidoalkylation reactions using BINOL-derived N-triflylphosphoramides as CPA catalysts is reported as a case of study. The model was implemented in a web server called MATEO: InterMolecular Amidoalkylation Theoretical Enantioselectivity Optimization, available online at: https://cptmltool.rnasa-imedir.com/CPTMLTools-Web/mateo . This new user-friendly online computational tool would enable sustainable optimization of reaction conditions that could lead to the design of new CPA catalysts along with new organic synthesis products.

Applied machine learning as a driver for polymeric biomaterials design

Polymers are ubiquitous to almost every aspect of modern society and their use in medical products is similarly pervasive. Despite this, the diversity in commercial polymers used in medicine is stunningly low. Considerable time and resources have been extended over the years towards the development of new polymeric biomaterials which address unmet needs left by the current generation of medical-grade polymers. Machine learning (ML) presents an unprecedented opportunity in this field to bypass the need for trial-and-error synthesis, thus reducing the time and resources invested into new discoveries critical for advancing medical treatments. Current efforts pioneering applied ML in polymer design have employed combinatorial and high throughput experimental design to address data availability concerns. However, the lack of available and standardized characterization of parameters relevant to medicine, including degradation time and biocompatibility, represents a nearly insurmountable obstacle to ML-aided design of biomaterials. Herein, we identify a gap at the intersection of applied ML and biomedical polymer design, highlight current works at this junction more broadly and provide an outlook on challenges and future directions.

Towards artificial intelligence-enabled extracellular vesicle precision drug delivery

Extracellular Vesicles (EVs), particularly exosomes, recently exploded into nanomedicine as an emerging drug delivery approach due to their superior biocompatibility, circulating stability, and bioavailability in vivo. However, EV heterogeneity makes molecular targeting precision a critical challenge. Deciphering key molecular drivers for controlling EV tissue targeting specificity is in great need. Artificial intelligence (AI) brings powerful prediction ability for guiding the rational design of engineered EVs in precision control for drug delivery. This review focuses on cutting-edge nano-delivery via integrating large-scale EV data with AI to develop AI-directed EV therapies and illuminate the clinical translation potential. We briefly review the current status of EVs in drug delivery, including the current frontier, limitations, and considerations to advance the field. Subsequently, we detail the future of AI in drug delivery and its impact on precision EV delivery. Our review discusses the current universal challenge of standardization and critical considerations when using AI combined with EVs for precision drug delivery. Finally, we will conclude this review with a perspective on future clinical translation led by a combined effort of AI and EV research.

Prediction of the Ibuprofen Loading Capacity of MOFs by Machine Learning

Metal-organic frameworks (MOFs) have been widely researched as drug delivery systems due to their intrinsic porous structures. Herein, machine learning (ML) technologies were applied for the screening of MOFs with high drug loading capacity. To achieve this, first, a comprehensive dataset was gathered, including 40 data points from more than 100 different publications. The organic linkers, metal ions, and the functional groups, as well as the surface area and the pore volume of the investigated MOFs, were chosen as the model’s inputs, and the output was the ibuprofen (IBU) loading capacity. Thereafter, various advanced and powerful machine learning algorithms, such as support vector regression (SVR), random forest (RF), adaptive boosting (AdaBoost), and categorical boosting (CatBoost), were employed to predict the ibuprofen loading capacity of MOFs. The coefficient of determination (R²) of 0.70, 0.72, 0.66, and 0.76 were obtained for the SVR, RF, AdaBoost, and CatBoost approaches, respectively. Among all the algorithms, CatBoost was the most reliable, exhibiting superior performance regarding the sparse matrices and categorical features. Shapley additive explanations (SHAP) analysis was employed to explore the impact of the eigenvalues of the model’s outputs. Our initial results indicate that this methodology is a well generalized, straightforward, and cost-effective method that can be applied not only for the prediction of IBU loading capacity, but also in many other biomaterials projects.

Prediction of Antileishmanial Compounds: General Model, Preparation, and Evaluation of 2-Acylpyrrole Derivatives

In this work, the SOFT.PTML tool has been used to pre-process a ChEMBL dataset of pre-clinical assays of antileishmanial compound candidates. A comparative study of different ML algorithms, such as logistic regression (LOGR), support vector machine (SVM), and random forests (RF), has shown that the IFPTML-LOGR model presents excellent values of specificity and sensitivity (81-98%) in training and validation series. The use of this software has been illustrated with a practical case study focused on a series of 28 derivatives of 2-acylpyrroles 5a,b, obtained through a Pd(II)-catalyzed C-H radical acylation of pyrroles. Their in vitro leishmanicidal activity against visceral (L. donovani) and cutaneous (L. amazonensis) leishmaniasis was evaluated finding that compounds 5bc (IC₅₀ = 30.87 μM, SI > 10.17) and 5bd (IC₅₀ = 16.87 μM, SI > 10.67) were approximately 6-fold more selective than the drug of reference (miltefosine) in in vitro assays against L. amazonensis promastigotes. In addition, most of the compounds showed low cytotoxicity, CC₅₀ > 100 μg/mL in J774 cells. Interestingly, the IFPMTL-LOGR model predicts correctly the relative biological activity of these series of acylpyrroles. A computational high-throughput screening (cHTS) study of 2-acylpyrroles 5a,b has been performed calculating >20,700 activity scores vs a large space of 647 assays involving multiple Leishmania species, cell lines, and potential target proteins. Overall, the study demonstrates that the SOFT.PTML all-in-one strategy is useful to obtain IFPTML models in a friendly interface making the work easier and faster than before. The present work also points to 2-acylpyrroles as new lead compounds worthy of further optimization as antileishmanial hits.

PTML Modeling for Pancreatic Cancer Research: In Silico Design of Simultaneous Multi-Protein and Multi-Cell Inhibitors

Pancreatic cancer (PANC) is a dangerous type of cancer that is a major cause of mortality worldwide and exhibits a remarkably poor prognosis. To date, discovering anti-PANC agents remains a very complex and expensive process. Computational approaches can accelerate the search for anti-PANC agents. We report for the first time two models that combined perturbation theory with machine learning via a multilayer perceptron network (PTML-MLP) to perform the virtual design and prediction of molecules that can simultaneously inhibit multiple PANC cell lines and PANC-related proteins, such as caspase-1, tumor necrosis factor-alpha (TNF-alpha), and the insulin-like growth factor 1 receptor (IGF1R). Both PTML-MLP models exhibited accuracies higher than 78%. Using the interpretation from one of the PTML-MLP models as a guideline, we extracted different molecular fragments desirable for the inhibition of the PANC cell lines and the aforementioned PANC-related proteins and then assembled some of those fragments to form three new molecules. The two PTML-MLP models predicted the designed molecules as potentially versatile anti-PANC agents through inhibition of the three PANC-related proteins and multiple PANC cell lines. Conclusions: This work opens new horizons for the application of the PTML modeling methodology to anticancer research.

Machine Learning-Driven Biomaterials Evolution

Biomaterials is an exciting and dynamic field, which uses a collection of diverse materials to achieve desired biological responses. While there is constant evolution and innovation in materials with time, biomaterials research has been hampered by the relatively long development period required. In recent years, driven by the need to accelerate materials development, the applications of machine learning in materials science has progressed in leaps and bounds. The combination of machine learning with high-throughput theoretical predictions and high-throughput experiments (HTE) has shifted the traditional Edisonian (trial and error) paradigm to a data-driven paradigm. In this review, each type of biomaterial and their key properties and use cases are systematically discussed, followed by how machine learning can be applied in the development and design process. The discussions are classified according to various types of materials used including polymers, metals, ceramics, and nanomaterials, and implants using additive manufacturing. Last, the current gaps and potential of machine learning to further aid biomaterials discovery and application are also discussed.

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.

Machine learning and materials modelling interpretation of in vivo toxicological response to TiO2 nanoparticles library (UV and non-UV exposure)

Assessing the risks of nanomaterials/nanoparticles (NMs/NPs) under various environmental conditions requires a more systematic approach, including the comparison of effects across many NMs with identified different but related characters/descriptors. Hence, there is an urgent need to provide coherent (eco)toxicological datasets containing comprehensive toxicity information relating to a diverse spectra of NPs characters. These datasets are test benches for developing holistic methodologies with broader applicability. In the present study we assessed the effects of a custom design Fe-doped TiO₂ NPs library, using the soil invertebrate Enchytraeus crypticus (Oligochaeta), via a 5-day pulse via aqueous exposure followed by a 21-days recovery period in soil (survival, reproduction assessment). Obviously, when testing TiO₂, realistic conditions should include UV exposure. The 11 Fe-TiO₂ library contains NPs of size range between 5-27 nm with varying %Fe (enabling the photoactivation of TiO₂ at energy wavelengths in the visible-light range). The NPs were each described by 122 descriptors, being a mixture of measured and atomistic model descriptors. The data were explored using single and univariate statistical methods, combined with machine learning and multiscale modelling techniques. An iterative pruning process was adopted for identifying automatically the most significant descriptors. TiO₂ NPs toxicity decreased when combined with UV. Notably, the short-term water exposure induced lasting biological responses even after longer-term recovery in clean exposure. The correspondence with Fe-content correlated with the band-gap hence the reduction of UV oxidative stress. The inclusion of both measured and modelled materials data benefitted the explanation of the results, when combined with machine learning.

High-throughput screening of nanoparticles in drug delivery

The use of pharmacologically active compounds to manage and treat diseases is of utmost relevance in clinical practice. It is well recognized that spatial-temporal control over the delivery of these biomolecules will greatly impact their pharmacokinetic profile and ultimately their therapeutic effect. Nanoparticles (NPs) prepared from different materials have been tested successfully in the clinic for the delivery of several biomolecules including non-coding RNAs (siRNA and miRNA) and mRNAs. Indeed, the recent success of mRNA vaccines is in part due to progress in the delivery systems (NP based) that have been developed for many years. In most cases, the identification of the best formulation was done by testing a small number of novel formulations or by modification of pre-existing ones. Unfortunately, this is a low throughput and time-consuming process that hinders the identification of formulations with the highest potential. Alternatively, high-throughput combinatorial design of NP libraries may allow the rapid identification of formulations with the required release and cell/tissue targeting profile for a given application. Combinatorial approaches offer several advantages over conventional methods since they allow the incorporation of multiple components with varied chemical properties into materials, such as polymers or lipid-like materials, that will subsequently form NPs by self-assembly or chemical conjugation processes. The current review highlights the impact of high-throughput in the development of more efficient drug delivery systems with enhanced targeting and release kinetics. It also describes the current challenges in this research area as well as future directions.

Biomaterials by design: Harnessing data for future development

Biomaterials is an interdisciplinary field of research to achieve desired biological responses from new materials, regardless of material type. There have been many exciting innovations in this discipline, but commercialization suffers from a lengthy discovery to product pipeline, with many failures along the way. Success can be greatly accelerated by harnessing machine learning techniques to comb through large amounts of data. There are many potential benefits of moving from an unstructured empirical approach to a development strategy that is entrenched in data. Here, we discuss the recent work on the use of machine learning in the discovery and design of biomaterials, including new polymeric, metallic, ceramics, and nanomaterials, and how machine learning can interface with emerging use cases of 3D printing. We discuss the steps for closer integration of machine learning to make this exciting possibility a reality.

Computational Drug Repurposing for Antituberculosis Therapy: Discovery of Multi-Strain Inhibitors

Tuberculosis remains the most afflicting infectious disease known by humankind, with one quarter of the population estimated to have it in the latent state. Discovering antituberculosis drugs is a challenging, complex, expensive, and time-consuming task. To overcome the substantial costs and accelerate drug discovery and development, drug repurposing has emerged as an attractive alternative to find new applications for “old” drugs and where computational approaches play an essential role by filtering the chemical space. This work reports the first multi-condition model based on quantitative structure–activity relationships and an ensemble of neural networks (mtc-QSAR-EL) for the virtual screening of potential antituberculosis agents able to act as multi-strain inhibitors. The mtc-QSAR-EL model exhibited an accuracy higher than 85%. A physicochemical and fragment-based structural interpretation of this model was provided, and a large dataset of agency-regulated chemicals was virtually screened, with the mtc-QSAR-EL model identifying already proven antituberculosis drugs while proposing chemicals with great potential to be experimentally repurposed as antituberculosis (multi-strain inhibitors) agents. Some of the most promising molecules identified by the mtc-QSAR-EL model as antituberculosis agents were also confirmed by another computational approach, supporting the capabilities of the mtc-QSAR-EL model as an efficient tool for computational drug repurposing.

Impact of atomistic or crystallographic descriptors for classification of gold nanoparticles

Machine learning models are known to be sensitive to the features used to train them, but there is currently no way to predict the impact of using different features prior to feature extraction. This is particularly important to fields such as nanotechnology that are highly multi-disciplinary, and samples can be characterised many different ways depending on the preferences of individual researchers. Does it matter if nanomaterials are described using the interatomic coordinations or more complex order parameters? In this study we compare results of supervised and unsupervised learning on a single set of gold nanoparticles that has been characterised by two different descriptors, each with a unique feature space. We find that there are some consistencies, and model selection is descriptor-agnostic, but the level of detail and the type of information that can be extracted from the results is sensitive to the way the particles are described. Unsupervised clustering revealed that an atomistic descriptor provides a finer-grained interpretation and clusters that are sub-clusters of a more sophisticated crystallographic descriptor, which is consistent with both how the features were calculated, and how they are interpreted in the domain. A supervised classifier revealed that the types of features responsible for the separation are related to the bulk structure, regardless of the descriptor, but capture different types of information. For both the atomistic and crystallographic descriptor the gradient boosting decision tree classifier gave superior results of F1-scores of 0.96 and 0.98, respectively, with excellent precision and recall, even though the clustering presented a challenging multi-classification problem.

Palladium-mediated synthesis and biological evaluation of C-10b substituted Dihydropyrrolo[1,2-b]isoquinolines as antileishmanial agents

The development of new molecules for the treatment of leishmaniasis is, a neglected parasitic disease, is urgent as current anti-leishmanial therapeutics are hampered by drug toxicity and resistance. The pyrrolo[1,2-b]isoquinoline core was selected as starting point, and palladium-catalyzed Heck-initiated cascade reactions were developed for the synthesis of a series of C-10 substituted derivatives. Their in vitro leishmanicidal activity against visceral (L. donovani) and cutaneous (L. amazonensis) leishmaniasis was evaluated. The best activity was found, in general, for the 10-arylmethyl substituted pyrroloisoquinolines. In particular, 2ad (IC₅₀ = 3.30 μM, SI > 77.01) and 2bb (IC₅₀ = 3.93 μM, SI > 58.77) were approximately 10-fold more potent and selective than the drug of reference (miltefosine), against L. amazonensis on in vitro promastigote assays, while 2ae was the more active compound in the in vitro amastigote assays (IC₅₀ = 33.59 μM, SI > 8.93). Notably, almost all compounds showed low cytotoxicity, CC₅₀ > 100 μg/mL in J774 cells, highest tested dose. In addition, we have developed the first Perturbation Theory Machine Learning (PTML) algorithm able to predict simultaneously multiple biological activity parameters (IC₅₀, K_i, etc.) vs. any Leishmania species and target protein, with high values of specificity (>98%) and sensitivity (>90%) in both training and validation series. Therefore, this model may be useful to reduce time and assay costs (material and human resources) in the drug discovery process.

IFPTML mapping of nanoparticle antibacterial activity vs. pathogen metabolic networks

Nanoparticles are useful antimicrobial drug-release systems, but some nanoparticles also exhibit antibacterial activity. However, investigation of their antibacterial activity is a difficult and slow process due to the numerous combinations of nanoparticle size, shape, and composition vs. biological tests, assay organisms, and multiple activity parameters to be measured. Additionally, the overuse of antibiotics has led to the emergence of resistant bacterial strains with different metabolic networks. Computational models may speed up this process, but the models reported to date do not to consider all the previous factors, and the data sources are dispersed and not curated. Thus, herein, we used an information fusion, perturbation-theory machine learning (IFPTML) approach, which is introduced by us for the first time, to fit a model for the discovery of antibacterial nanoparticles. The dataset studied had 15 classes of nanoparticles (1-100 nm) with most cases in the range of 1-50 nm vs. >20 pathogenic bacteria species with different metabolic networks. The nanoparticles studied included metal nanoparticles of Au, Ag, and Cu; oxide nanoparticles of Zn, Cu, La, Al, Fe, Sn, Ti, Cd, and Si; and metal salt nanoparticles of CuI and CdS. We used the SOFT.PTML software (our own application) with a user-friendly interface for the IFPTML calculations and a control statistics package. Using SOFT.PTML, we found a linear logistic regression equation that could model 4 biological activity parameters using only 8 variables with χ2 = 2265.75, p-level <0.05, sensitivity, Sn = 79.4, and specificity, Sp = 99.3, for 3213 cases (nanoparticle-bacteria pairs) in the training series. The model had Sn = 80.8 and Sp = 99.3 for 2114 cases in the external validation series. We also developed a random forest non-linear model with higher values of Sn and Sp = 98-99% in the training/validation series, although it was more complicated to use. SOFT.PTML has been demonstrated to be a useful tool for the analysis of complex data in nanotechnology. We also introduced a new anabolism-catabolism unbalance index of metabolic networks to reveal the biological connotation of the IFPTML predictions for antibacterial nanoparticles. These new models open a new door for the discovery of NPs vs. new bacterial species and strains with different topological structures of their metabolic networks.

Computational and Experimental Approaches to Investigate Lipid Nanoparticles as Drug and Gene Delivery Systems

Lipid nanoparticles (LNPs) have been widely applied in drug and gene delivery. More than twenty years ago, DoxilTM was the first LNPs-based drug approved by the US Food and Drug Administration (FDA). Since then, with decades of research and development, more and more LNP-based therapeutics have been used to treat diverse diseases, which often offer the benefits of reduced toxicity and/or enhanced efficacy compared to the active ingredients alone. Here, we provide a review of recent advances in the development of efficient and robust LNPs for drug/gene delivery. We emphasize the importance of rationally combining experimental and computational approaches, especially those providing multiscale structural and functional information of LNPs, to the design of novel and powerful LNP-based delivery systems.