BindingSite-AugmentedDTA: enabling a next-generation pipeline for interpretable prediction models in drug repurposing.

While research into drug-target interaction (DTI) prediction is fairly mature, generalizability and interpretability are not always addressed in the existing works in this field. In this paper, we propose a deep learning (DL)-based framework, called BindingSite-AugmentedDTA, which improves drug-target affinity (DTA) predictions by reducing the search space of potential-binding sites of the protein, thus making the binding affinity prediction more efficient and accurate. Our BindingSite-AugmentedDTA is highly generalizable as it can be integrated with any DL-based regression model, while it significantly improves their prediction performance. Also, unlike many existing models, our model is highly interpretable due to its architecture and self-attention mechanism, which can provide a deeper understanding of its underlying prediction mechanism by mapping attention weights back to protein-binding sites. The computational results confirm that our framework can enhance the prediction performance of seven state-of-the-art DTA prediction algorithms in terms of four widely used evaluation metrics, including concordance index, mean squared error, modified squared correlation coefficient ($r^2_m$) and the area under the precision curve. We also contribute to three benchmark drug-traget interaction datasets by including additional information on 3D structure of all proteins contained in those datasets, which include the two most commonly used datasets, namely Kiba and Davis, as well as the data from IDG-DREAM drug-kinase binding prediction challenge. Furthermore, we experimentally validate the practical potential of our proposed framework through in-lab experiments. The relatively high agreement between computationally predicted and experimentally observed binding interactions supports the potential of our framework as the next-generation pipeline for prediction models in drug repurposing.

How artificial intelligence may help the Covid-19 pandemic: Pitfalls and lessons for the future.

The clinical severity, rapid transmission and human losses due to coronavirus disease 2019 (Covid-19) have led the World Health Organization to declare it a pandemic. Traditional epidemiological tools are being significantly complemented by recent innovations especially using artificial intelligence (AI) and machine learning. AI-based model systems could improve pattern recognition of disease spread in populations and predictions of outbreaks in different geographical locations. A variable and a minimal amount of data are available for the signs and symptoms of Covid-19, allowing a composite of maximum likelihood algorithms to be employed to enhance the accuracy of disease diagnosis and to identify potential drugs. AI-based forecasting and predictions are expected to complement traditional approaches by helping public health officials to select better response and preparedness measures against Covid-19 cases. AI-based approaches have helped address the key issues but a significant impact on the global healthcare industry is yet to be achieved. The capability of AI to address the challenges may make it a key player in the operation of healthcare systems in future. Here, we present an overview of the prospective applications of the AI model systems in healthcare settings during the ongoing Covid-19 pandemic.

Injecting drug use & HIV prevalence among female sex workers: Evidence from the National Integrated Biological & Behavioural Surveillance, India.

Background & objectives: Female sex workers (FSWs) who inject drugs (FSW-IDs) have a higher risk of HIV infection and transmission. Understanding the socio-demographic characteristics and other risk behaviours among FSW-IDs will help in strengthening targeted interventions for HIV prevention and management. In the present study, the HIV prevalence, associated socio-demographic characteristics and risk behaviours among FSWs who injected drugs (FSW-IDs) and those who did not ID (FSW-NIDs) was determined in India. Methods: The national cross-sectional, community-based, integrated biological and behavioural surveillance was conducted in 2014-2015 at 73 randomly selected FSW domains across 28 States and Union Territories in India. The sample size was fixed at 400 for each domain, and a probability-based sampling method was followed. The data were analyzed by logistic regression methods. Results: Data from 27,007 FSWs were included in the analysis, of which 802 (3%) were FSW-IDs. HIV prevalence among FSW-IDs was significantly higher than that in FSW-NIDs (4.5 vs. 1.9%). Univariate analysis showed that factors significantly associated with higher HIV prevalence among FSW-IDs were older age, sex work as the only source of income, dissolved marriage, living with a sex worker, urban locality of sex work and consumption of alcohol or oral drugs. In multivariable analysis, factors such as older age of FSW-IDs (35 yr and above), having a dissolved marriage and sex work being the only source of income were observed to be independently and significantly associated with higher HIV prevalence. Interpretation & conclusions: Scaling up the HIV preventive interventions for FSW-IDs, such as facilitating awareness and improved access to needle and syringe exchange programme (NSEP) and opioid substitution therapy (OST), encouraging safe sex and injecting practices, educating on the harmful effects of alcohol and drugs and providing alternative vocation options to secure their financial needs are several strategies that may reduce HIV transmission among FSWs.

Modeling Oxidation of AlCoCrFeNi High-Entropy Alloy Using Stochastic Cellular Automata.

Together with the thermodynamics and kinetics, the complex microstructure of high-entropy alloys (HEAs) exerts a significant influence on the associated oxidation mechanisms in these concentrated solid solutions. To describe the surface oxidation in AlCoCrFeNi HEA, we employed a stochastic cellular automata model that replicates the mesoscale structures that form. The model benefits from diffusion coefficients of the principal elements through the native oxides predicted by using molecular simulations. Through our examination of the oxidation behavior as a function of the alloy composition, we corroborated that the oxide scale growth is a function of the complex chemistry and resultant microstructures. The effect of heat treatment on these alloys is also simulated by using reconstructed experimental micrographs. When they are in a single-crystal structure, no segregation is noted for α-Al2O3 and Cr2O3, which are the primary scale-forming oxides. However, a coexistent separation between Al2O3 and Cr2O3 oxide scales with the Al-Ni- and Cr-Fe-rich regions is predicted when phase-separated microstructures are incorporated into the model.

Transfer Learned Designer Polymers For Organic Solar Cells.

Organic photovoltaic (OPV) materials have been examined extensively over the past two decades for solar cell applications because of the potential for device flexibility, low-temperature solution processability, and negligible environmental impact. However, discovery of new candidate OPV materials, especially polymer-based electron donors, that demonstrate notable power conversion efficiencies (PCEs), is nontrivial and time-intensive exercise given the extensive set of possible chemistries. Recent progress in machine learning accelerated materials discovery has facilitated to address this challenge, with molecular line representations, such as Simplified Molecular-Input Line-Entry Systems (SMILES), gaining popularity as molecular fingerprints describing the donor chemical structures. Here, we employ a transfer learning based recurrent neural (LSTM) model, which harnesses the SMILES molecular fingerprints as an input to generate novel designer chemistries for OPV devices. The generative model, perfected on a small focused OPV data set, predicts new polymer repeat units with potentially high PCE. Calculations of the similarity coefficient between the known and the generated polymers corroborate the accuracy of the model predictability as a function of the underlying chemical specificity. The data-enabled framework is sufficiently generic for use in accelerated machine learned materials discovery for various chemistries and applications, mining the hitherto available experimental and computational data.

Elasto-morphology of P3HT:PCBM bulk heterojunction organic solar cells.

Predicting the mechanical properties of organic semiconductors is important when using these materials in flexible electronics applications. For instance, knowledge of the mechanical and thermal stability of thin film organic solar cells (OSCs) is critical for the roll-to-roll production of photovoltaic devices and their use under various operating conditions. Here, we examine the thermal and elasto-mechanical properties of the conjugated donor polymer poly-(3-hexylthiophene) (P3HT) and the interpenetrating mixtures of P3HT and phenyl-C61-butyric acid methyl (PCBM) ester bulk heterojunction (BHJ) active layers under the application of unidirectional tensile deformation using coarse-grained molecular dynamics (CGMD) simulations. The predictions are validated against previous experimental reports as well as with earlier modeling results derived using different intermolecular force fields. Our results reveal that PCBM molecules behave as anti-plasticizers when mixed with P3HT and tend to increase the tensile modulus and glass transition temperature, while decreasing the crack-onset strain relative to pure P3HT. The variations in the mechanical properties with the composition of the BHJ active layer suggest that, in the presence of small oligomers as additives in the BHJ, the P3HT:PCBM mixture resists the anti-plasticizing effect of PCBM molecules due to the low tensile modulus of the short polymer chains.

Nonclinical safety evaluation of erenumab, a CGRP receptor inhibitor for the prevention of migraine.

Calcitonin gene-related peptide (CGRP) and its receptor have been implicated as a key mediator in the pathophysiology of migraine. Thus, erenumab, a monoclonal antibody antagonist of the CGRP receptor, administered as a once monthly dose of 70 or 140 mg has been approved for the preventive treatment of migraine in adults. Due to the species specificity of erenumab, the cynomolgus monkey was used in the pharmacology, pharmacokinetics, and toxicology studies to support the clinical program. There were no effects of erenumab on platelets in vitro (by binding, activation or phagocytosis assays). Specific staining of human tissues with erenumab did not indicated any off-target binding. There were no erenumab-related findings in a cardiovascular safety pharmacology study in cynomolgus monkeys or in vitro in human isolated coronary arteries. Repeat-dose toxicology studies conducted in cynomolgus monkeys at dose levels up to 225 mg/kg (1 month) or up to 150 mg/kg (up to 6 months) with twice weekly subcutaneous (SC) doses showed no evidence of erenumab-mediated adverse toxicity. There were no effects on pregnancy, embryo-fetal or postnatal growth and development in an enhanced pre-postnatal development study in the cynomolgus monkey. There was evidence of placental transfer of erenumab based on measurable serum concentrations in the infants up to 3 months post birth. The maternal and developmental no-observed-effect level (NOEL) was the highest dose tested (50 mg/kg SC Q2W). These nonclinical data in total indicate no safety signal of concern to date and provide adequate margins of exposure between the observed safe doses in animals and clinical dose levels.

Reduced Thermal Transport in the Graphene/MoS2/Graphene Heterostructure: A Comparison with Freestanding Monolayers.

The thermal conductivity of the graphene-encapsulated MoS2 (graphene/MoS2/graphene) van der Waals heterostructure is determined along the armchair and zigzag directions with different twist angles between the layers using molecular dynamics (MD) simulations. The differences in the predictions relative to those of the monolayers are analyzed using the phonon power spectrum and phonon lifetimes obtained by spectral energy density analysis. The thermal conductivity of the heterostructure is predominantly isotropic. The out-of-plane phonons of graphene are suppressed because of the interaction between the adjacent layers that results in the reduced phonon lifetime and thermal conductivity relative to monolayer graphene. The occurrence of an additional nonzero phonon branch at the Γ point in the phonon dispersion curves of the heterostructure corresponds to the breathing modes resulting from stacking of the layers in the heterostructure. The thermal sheet conductance of the heterostructure being an order of magnitude larger than that of monolayer MoS2, this van der Waals material is potentially suitable for efficient thermal packaging of photoelectronic devices. The interfacial thermal conductance of the graphene/MoS2 bilayer as a function of the heat flow direction shows weak thermal rectification.

Azobenzene switch with a long-lived cis-state to photocontrol the enzyme activity of a histone deacetylase-like amidohydrolase.

The control of enzymes by use of an external stimulus such as light enables the temporal and spatial regulation of defined chemical reactions in a highly precise manner. In this work we investigated and characterized the reversible photocontrol of a bacterial histone deacetylase-like amidohydrolase (HDAH) from Bordetella/Alcaligenes strain FB188, which holds great potential to control deacetylation reactions of a broad spectrum of substrates in biotechnological and biomedical applications. Several HDAH variants with a single surface accessible cysteine close to the active site were developed and covalently modified by a monofunctional azobenzene-based photoswitch [4-phenylazomaleinanil (4-PAM)]. The enzymatic activity of three HDAH variants (M30C, S20C and M150C) were shown to be controlled by light. The thermal cis-to-trans relaxation of azobenzene conjugated to HDAH was up to 50-fold retarded compared to unbound 4-PAM allowing light pulse switching rather than continuing irradiation to maintain the thermodynamically less stable cis-state of covalently attached 4-PAM.

Influence of longitudinal isotope substitution on the thermal conductivity of carbon nanotubes: results of nonequilibrium molecular dynamics and local density functional calculations.

We report reverse nonequilibrium molecular dynamics calculations of the thermal conductivity of isotope substituted (10,10) carbon nanotubes (CNTs) at 300 K. (12)C and (14)C isotopes both at 50% content were arranged either randomly, in bands running parallel to the main axis of the CNTs or in bands perpendicular to this axis. It is found that the systems with randomly distributed isotopes yield significantly reduced thermal conductivity. In contrast, the systems where the isotopes are organized in patterns parallel to the CNTs axis feature no reduction in thermal conductivity when compared with the pure (14)C system. Moreover, a reduction of approximately 30% is observed in the system with the bands of isotopes running perpendicular to the CNT axis. The computation of phonon dispersion curves in the local density approximation and classical densities of vibrational states reveal that the phonon structure of carbon nanotubes is conserved in the isotope substituted systems with the ordered patterns, yielding high thermal conductivities in spite of the mass heterogeneity. In order to complement our conclusions on the (12)C-(14)C mixtures, we computed the thermal conductivity of systems where the (14)C isotope was turned into pseudo-atoms of 20 and 40 atomic mass units.