Abstract

Drug-Drug Interactions (DDIs) often cause significant adverse
events, making early identification crucial, especially with rising
polypharmacy. While most methods focus on binary prediction, multiclass
DDI prediction is more advantageous as it identifies specific interaction
types. However, this is challenging due to class imbalance, with
many rare interaction types compared to dominant ones. We propose
a multiclass DDI prediction framework comprising three components:
Stage 1 involves feature extraction, where the molecular structures of
two drugs are modeled as graphs using a graph attention network (GAT)
as an encoder, then combined with Drug Repurposing Knowledge Graph
(DRKG) embeddings to produce enhanced drug features. Stage 2 involves
expert modelling, where the paired drug representations are processed by
two Mamba State Space Model (SSM)- based experts: a standard Mamba
expert and a class-reweighted Mamba expert, which efficiently capture
sequential dependencies and interaction patterns between the drug representations.
Stage 3 involves gated fusion, where a confidence-based gating
strategy integrates the two experts to generate final predictions. The
performance on DrugBank and DDIMDL benchmark datasets achieved
macro F1 scores of 90.68% and 83.99%, demonstrating our framework's
effectiveness in multiclass DDI prediction. The source code is publicly
available at https://github.com/devalab/MambaDDI.

Abstract

Cardiovascular diseases continue to be a global health challenge, and early detection of congestive heart failure requires innovative diagnostic approaches. This study introduces a wearable, noninvasive bio-impedance-based monitoring system that addresses critical limitations in using bio-impedance for thoracic fluid detection. The novel design of the modular electrode belt, featuring a unique sliding electrode configuration with precision-scaled markings and adjustable positioning, represents a significant advancement in biomedical device design. Unlike previous wearable systems that struggled with inconsistent electrode placement, our belt incorporates a robust mechanism allowing precise alignment across different chest sizes, effectively mitigating the long-standing challenge of sensor reading variations due to electrode repositioning. Along with the hardware innovation, we present a Graph Neural Network architecture that uniquely represents each individual's data collected using our device as a graph, capturing complex interactions between electrode distances and their respective impedance values at multiple frequencies. The model employs an edge attention mechanism and graph attention convolution layer, which enables nuanced interpretation of bio-impedance signals across multiple thoracic regions. Achieving a Mean Squared Error of 0.8675, our approach demonstrates superior performance in predicting thoracic fluid percentage. Our research offers a comprehensive, noninvasive solution for thoracic fluid measurement, which can lead to early disease detection and patient care.

Abstract

Determining spin state energy gaps (SSE) of 3d transition metal complexes (TMCs) is a major challenge in theoretical chemistry, as high-level quantum methods, though reliable, are computationally impractical for large-scale studies. This work explores a machine learning (ML)-based approach to predict DFT adiabatic SSE gaps using descriptors derived from a single high-spin DFT calculation. This approach is adopted to eliminate the differential treatment of electronic correlation between high-spin and low-spin structures. Our descriptors aim to incorporate the knowledge of crystal field theory into the ML model. They include atomic energy levels of bare metal ions, natural charges of ligating atoms, d-orbital molecular orbital eigenvalues derived from an high spin calculation, HOMO-LUMO gaps of free ligands, and simple identity-based features. We train ML models on 1434 SSE values spanning 934 complexes and demonstrate their transferability to more challenging complexes having bidentate π-bonding ligands despite being trained on simpler Werner-type monodentate complexes. We achieved a minimum MAE of 4.0 kcal mol−1 on the monodentate test set, and maintained a comparable MAE of 6.6 kcal mol−1 in the transferability assessment. This approach bypasses the need for multi-reference low-spin optimizations while retaining predictive accuracy, offering a cost-effective strategy for SSE estimation in transition metal chemistry. We hope the insights covered in this study will contribute to the development of additional electronic structure-based descriptors for SSE predictions.

Abstract

Density functional theory comparison of the role of Au and Cu on the catalysis of aerobic oxidation of benzyl alcohol to benzaldehyde was performed. Four metal clusters were analyzed, namely pure Au13 and Cu13 and doped Au12 Cu and Cu12 Au clusters. This study focuses on the previously reported minimum energy pathway for oxidizing two molecules of benzyl alcohol per adsorbed O2 . The reaction energy pathway diverges at the benzylic H abstraction step. For catalysts rich in Au, dissociation of O-O dissociation before this step is not required, conversely for Cu rich catalysts, O-O dissociation occurs prior to benzylic H abstraction step. The first transition state for the abstraction of alcoholic hydrogen by O2 presents a minimal energy barrier on the Au13 catalyst, while it is absent on the other three catalysts. The calculations exhibit that the reaction rate is primarily influenced by the adsorption of reactants and desorption of products, aligning with experimental findings. In Cu rich clusters, the adsorption of O2 is more favorable due to effective charge transfer between the cluster and O2 resulting in high adsorption energy. Additionally, the desorption of products is significantly endothermic on Cu rich catalysts, whereas it is exothermic on Au rich catalysts.

Abstract

Recent progress and development of artificial intelligence and machine learning (AI/ML) techniques have enabled addressing complex biomolecular problems. AI/ML models learn the underlying distribution of data they are trained on and when exposed to new inputs, they make predictions based on patterns and relationships previously observed in the training set. Further, generative artificial intelligence (GenAI) can be used to accurately generate protein structure or sequence from specific selected properties. This review specifically focuses on the applications of AI/ML in predicting important functional properties of proteins, and the potential prospects of reverse-engineering in depicting the sequence and structure, from available protein-property information.

Abstract

Exploring promiscuous catalytic activity of enzymes to stereoselective transformations uncovers ecofriendly asymmetric catalysts and provides scope to execute new-to-nature chemistry. Biocatalytic promiscuous diastereoselective Henry reaction (DHR) is restricted to a few hydroxynitrile lyases (HNLs), often engineered, with narrow synthetic scope and anti-selectivity. Here, we report a single native enzyme exhibiting complementary diastereoselectivity in an asymmetric Henry reaction. Baliospermum montanum HNL (BmHNL)-catalyzed DHR enabled the production of thirty-two anti-(1S,2R)-β-nitroalcohols with up to >99% conversion, >99% ee, and >99% de using longer nitroalkanes. It afforded syn diastereomers, i.e., (1S,2S)-β-nitroalcohols as thermodynamically controlled products by manipulating the reaction conditions. Reuse of BmHNL in DHR for 20 cycles has empowered the synthesis of (1S,2R)-2-nitro-1-phenylpropan-1-ol (NPP), where it retained 82% of its initial activity, maintained >99% ee on each cycle, and provided total turnover number >3.7 × 104, which is 140-fold higher compared to the only existing biocatalytic DHR. Our study demonstrates a gram-scale synthesis of (1S,2R)-NPP and a preparative-scale synthesis of (1S,2S)-NPP, a precursor to d-norpseudoephedrine used for the treatment of obesity. The origin of the inverse diastereoselectivity was probed using isotope labeling studies. Molecular modeling of the reaction using density functional theory methods reveals the underlying mechanism of the diastereoselectivity and the competing kinetic vs thermodynamic nature of the reaction. Despite a promiscuous catalytic activity, the outstanding catalytic efficiency, broad synthetic scope, and complementary diastereoselectivity of the native enzyme of BmHNL on DHR are remarkable, which illustrates it as a specialized biocatalyst for greener technology and stereocontrolled production of diverse β-nitroalcohols.

Abstract

AuCu nanoclusters have widespread application in reactions like activation of CO2, selective oxidation, and cross-coupling reactions. In this study, we investigate the stepwise doping of copper atoms in a pure 13-atom gold cluster, denoted as AumCun (m+n = 13). The genetic algorithm based on the artificial bee colony algorithm has been utilized to model various isomers of each composition. The potential energy landscape of these clusters was analyzed by means of the density functional theory method with pure Perdew−Burke−Ernzerhof (PBE) functional. We identify the minimum energy isomer for each cluster composition to evaluate molecular properties like HOMO-LUMO gap, binding energy/atom, second order difference in energy, vertical ionization energy, and vertical electron affinity. Notably, the introduction of copper atoms in these clusters enhances their stability and reactivity. Distinct odd-even oscillations due to close shell electronic configurations are absent, as all cluster compositions have an overall open shell configuration. To assess the catalytic activity of the clusters, we study the adsorption energies of small molecules like O 2 and C 2 H 4 on all available sites
on the cluster. This study thereby comprehensively explores the range of copper-doped 13-atom gold cluster compositions and their implications on their structure-property relationships vital for catalysis and nanomaterial applications.

Abstract

Bioimpedance is an extensively studied non-invasive
technique with diverse applications in biomedicine. This comprehensive review delves into the foundational concepts, technical
intricacies, and practical implementations of bioimpedance. It
elucidates the underlying principles governing bioimpedance
measurements, including the relevant physics equations employed
for estimating body fluid levels. Moreover, a thorough examination of the prevalent single-chip analog front end (AFE)
available in the market, such as AD5933, MAX30001, AD5940,
and AFE4300, is conducted, shedding light on their specifications
and functionalities. The review focuses on using bioimpedance to
assess thoracic impedance for heart failure detection by utilizing
the relation between lung water and heart failure. Traditional
techniques are compared with bioimpedance-based methods,
demonstrating the latter's efficacy as a non-invasive tool for
cardiac evaluation. In addition, the review addresses the technical limitations and challenges associated with bioimpedance.
Pertinent issues such as contact impedance, motion artifacts, calibration, and validation regarding their impact on measurement
precision and dependability are thoroughly examined. The review
also explores strategies and advancements in using artificial
intelligence to mitigate these challenges.

Abstract

Chemical reaction prediction, encompassing forward synthesis and retrosynthesis, stands as a fundamental challenge in organic synthesis. A widely adopted computational approach frames synthesis prediction as a sequence-to-sequence translation task, using the commonly used SMILES representation for molecules. The current evaluation of machine learning methods for retrosynthesis assumes perfect training data, overlooking imperfections in reaction equations in popular datasets, such as missing reactants, products, other physical and practical constraints such as temperature and cost, primarily due to a focus on the target molecule. This limitation leads to an incomplete representation of viable synthetic routes, especially when multiple sets of reactants can yield a given desired product. In response to these shortcomings, this study examines the prevailing evaluation methods and introduces comprehensive metrics designed to address imperfections in the dataset. Our novel metrics not only assess absolute accuracy by comparing predicted outputs with ground truth but also introduce a nuanced evaluation approach. We provide scores for partial correctness and compute adjusted accuracy through graph matching, acknowledging the inherent complexities of retrosynthetic pathways. Additionally, we explore the impact of small molecular augmentations while preserving chemical properties and employ similarity matching to enhance the assessment of prediction quality. We introduce SynFormer, a sequence-to-sequence model tailored for SMILES representation. It incorporates architectural enhancements to the original transformer, effectively tackling the challenges of chemical reaction prediction. SynFormer achieves a Top-1 accuracy of 53.2% on the USPTO-50k dataset, matching the performance of widely accepted models like Chemformer, but with greater efficiency by eliminating the need for pre-training.

Abstract

Drug-Drug Interactions (DDI) can trigger unexpected pharmacological consequences, including adverse drug events (ADE). The rise in polypharmacy underscores the importance of understanding how different drug molecules influence each other's pharmacological activities and necessitates the investigation of potential interactions between newly developed drugs and existing medications. Traditional laboratory methods for DDI detection are time-consuming, making the development of computational prediction methods crucial. This study introduces GraphDDI, a machine learning method utilizing Graph Neural Networks (GNN) to predict DDI accurately. The proposed methodology, trained end-to-end, comprises three stages. (1) Featurization stage: A GNN extracts atomic features from two drugs separately. (2) Interaction stage: an interaction map is calculated between all atom pairs of the drugs. (3) Prediction stage: the model combines the interaction map and drug features to create a unified representation of the drug molecules. Subsequently, the model concatenates these representations and employs a feed-forward neural network to predict the DDI. We demonstrate the efficacy of our proposed model in predicting both the presence of DDI and the specific types of interactions (DDI events). Comparative analysis reveals that our framework surpasses existing models, achieving an F1 score of 0.98 in predicting the existence of drug-drug interactions and 0.90 in categorizing DDI event types. The code is available in our GitHub repository (https://github.com/devalab/GraphDDI).