Efficient Discovery Research Articles

ECDANs: Efficient Temporal Causal Discovery from Autocorrelated and Non-stationary Data (Student Abstract)

eCDANs: Efficient Temporal Causal Discovery from Autocorrelated and Non-stationary Data (Student Abstract)

Hecogenin a Plant Derived Small Molecule as an Antagonist to BACE-1: A Potential Target for Neurodegenerative Disorders.

The field of drug discovery has recognized the significance of computer-aided drug design. Recent advancements in structure identification and characterization, bio-computational science and molecular biology have significantly contributed to the development of novel treatments for various diseases. Alzheimer's disease is prevalent in over 50 million affected people, with the pathological condition of amyloidal plaque formation by the beta-amyloidal peptide that results in lesions of the patient's brain, thus making the target prediction and treatment a hurdle. In this study, we evaluated the potential of 54 bioactive compounds from Justicia adhatoda L. and Sida cordifolia L. identified through LC-MS/MS against the β-site amyloid precursor cleaving enzyme (beta-secretase) that results in the formation of amyloidal plaques. To study the drug-likeness of the phytocompounds, Lipinski's rule of five for ADME profiling and toxicity prediction was performed. Molecular docking was performed using auto-dock tool of PyRx software; molecular dynamic simulations were performed using the Schrodinger suite. Molecular docking against BACE-1 protein revealed that hecogenin, identified from S. cordifolia has a broad spectrum of pharmacological applications and a binding affinity score of -11.3 kcal/Mol. The Hecogenin-BACE-1 protein complex was found to be stable after 30 ns of MD simulation, resulting in its substantial stability. Further studies focusing on the in vivo neuroprotective activity of hecogenin against the disease will pave the way for efficient drug discovery from natural sources in a precise manner.

Multi-Drug Transfer Learning Models for Efficient Pre-clinical Drug Discovery

Multi-Drug Transfer Learning Models for Efficient Pre-clinical Drug Discovery

Contrastive learning in protein language space predicts interactions between drugs and protein targets

Sequence-based prediction of drug-target interactions has the potential to accelerate drug discovery by complementing experimental screens. Such computational prediction needs to be generalizable and scalable while remaining sensitive to subtle variations in the inputs. However, current computational techniques fail to simultaneously meet these goals, often sacrificing performance of one to achieve the others. We develop a deep learning model, ConPLex, successfully leveraging the advances in pretrained protein language models ("PLex") and employing a protein-anchored contrastive coembedding ("Con") to outperform state-of-the-art approaches. ConPLex achieves high accuracy, broad adaptivity to unseen data, and specificity against decoy compounds. It makes predictions of binding based on the distance between learned representations, enabling predictions at the scale of massive compound libraries and the human proteome. Experimental testing of 19 kinase-drug interaction predictions validated 12 interactions, including four with subnanomolar affinity, plus a strongly binding EPHB1 inhibitor (KD = 1.3 nM). Furthermore, ConPLex embeddings are interpretable, which enables us to visualize the drug-target embedding space and use embeddings to characterize the function of human cell-surface proteins. We anticipate that ConPLex will facilitate efficient drug discovery by making highly sensitive in silico drug screening feasible at the genome scale. ConPLex is available open source at https://ConPLex.csail.mit.edu.

Probing phase stability in CrMoNbV using cluster expansion method, CALPHAD calculations and experiments

High entropy alloys (HEA), a novel class of materials with multiple principal elements, allow us to tune the properties within a wide composition range where the solid-solution phase is stable. However, it is difficult to experimentally determine the boundaries of the single-phase region. In this paper, we calculate the phase diagram of the CrMoNbV quaternary system, as well as its constituent binary and ternary subsystems, using a combination of the cluster expansion method and CALPHAD calculations. We further verify these results by carrying out experiments at two different compositions of the full quaternary system. Our work shows how these computational tools can enable efficient discovery and development of new HEAs.

DeepJoin: Joinable Table Discovery with Pre-Trained Language Models

Due to the usefulness in data enrichment for data analysis tasks, joinable table discovery has become an important operation in data lake management. Existing approaches target equi-joins, the most common way of combining tables for creating a unified view, or semantic joins, which tolerate misspellings and different formats to deliver more join results. They are either exact solutions whose running time is linear in the sizes of query column and target table repository, or approximate solutions lacking precision. In this paper, we propose DeepJoin, a deep learning model for accurate and efficient joinable table discovery. Our solution is an embedding-based retrieval, which employs a pre-trained language model (PLM) and is designed as one framework serving both equi- and semantic (with a similarity condition on word embeddings) joins for textual attributes with fairly small cardinalities. We propose a set of contextualization options to transform column contents to a text sequence. The PLM reads the sequence and is fine-tuned to embed columns to vectors such that columns are expected to be joinable if they are close to each other in the vector space. Since the output of the PLM is fixed in length, the subsequent search procedure becomes independent of the column size. With a state-of-the-art approximate nearest neighbor search algorithm, the search time is sublinear in the repository size. To train the model, we devise the techniques for preparing training data as well as data augmentation. The experiments on real datasets demonstrate that by training on a small subset of a corpus, DeepJoin generalizes to large datasets and its precision consistently outperforms other approximate solutions'. DeepJoin is even more accurate than an exact solution to semantic joins when evaluated with labels from experts. Moreover, when equipped with a GPU, DeepJoin is up to two orders of magnitude faster than existing solutions.

Accelerated Electrosynthesis Development Enabled by High-Throughput Experimentation

AbstractElectrochemical synthesis has recently emerged as an environmentally benign method for synthesizing value-added fine chemicals. Its unique reactivity has attracted significant interests of synthetic chemists to develop new redox chemistries. However, compared to conventional chemistry, the increased complexity caused by electrode materials, supporting electrolytes, and setup configurations create obstacles for efficient reaction discovery and optimization. The recent increasing adoption of high-throughput experimentation (HTE) in synthetic chemistry significantly expedites the synthesis development. Considering the potential of implementing HTE in electrosynthesis to tackle the challenges of increased parameter space, this short review aims at providing recent advances in the HTE technology for electrosynthesis, including electrocatalysts screening, device miniaturization, electroanalytical methods, artificial intelligence, and system integration. The discussed contents also cover some topics in HTE electrochemistry for areas other than synthetic chemistry, hoping to spark some inspirations for readers to use interdisciplinary techniques to solve challenges in synthetic electrochemistry.1 Introduction2 Parallelized Reaction Screening3 High-Throughput Screening for Electrocatalysts4 Miniaturization of Screening Devices5 Analytical Methods for Electrosynthesis Screening6 Artificial Intelligence for High-Throughput Screening7 Integrated Screening Systems8 Conclusion and Outlook

Towards the efficient discovery of meaningful functional dependencies

We propose the first framework for discovering the set of meaningful functional dependencies from data. This set contains the true positives among the set of functional dependencies that hold on the given data. Based on new data structures and original techniques for the dynamic computation of stripped partitions, we devise a new hybridization strategy that results in the first algorithm that can explore trade-offs between runtime efficiency and main memory usage. Using real-world benchmark data, we demonstrate that our algorithm outperforms the previous state-of-the-art in terms of runtime efficiency, and scalability in the number of rows and columns. We propose the number of redundant data values for ranking the functional dependencies that have been discovered. Our ranking helps separate false from true positives for applications, such as schema design. The remaining meaningful functional dependencies consist of the false negatives, that is, those functional dependencies that are only violated by the given data due to data inconsistency. We propose the computation of informative Armstrong relations to draw the attention of users to violations of functional dependencies that are meaningful for some application. We order the pairs of records in Armstrong relations based on the amount of inconsistency and redundancy caused by the associated functional dependencies, thereby pointing the attention to those most likely to be meaningful. As we demonstrate, these samples help separate false from true negatives, their perfect recall of meaningful functional dependencies can lead to a more complete acquisition of requirements and identification of dirty data, and may be computed faster than covers of functional dependencies. In addition, we demonstrate for the first time that non-redundant covers can offer a representation of functional dependencies that is much smaller than left-hand side reduced covers used in previous work. Such a compact representation of the output is easier to understand and explore by humans. We report all our results for different interpretations of missing values.

Blockchain Private Pools and Price Discovery

In this paper, we examine the balance between allocative efficiency and price discovery in blockchain-based finance. Our analysis reveals that the use of private pools, while improving blockspace allocation efficiency, negatively impacts price discovery. We show that the presence of private pools limits the amount of information revealed through pending orders and increases the execution risk of informed orders. We argue that price discovery can be improved by increasing the number of validators monitoring private pools or the block production rate on the chain.

Molecular networking as a natural products discovery strategy

The rapid development of bioinformatics tools has recently broken through the bottleneck in natural products research. These advances have enabled natural products researchers to rapidly separate and efficiently target and discover previously undescribed molecules. Among these advances, tandem mass spectrometry molecular networking is a promising method for rapidly de-replicating complex natural mixtures, thus leading to an accelerated revolution in the “art of natural products isolation” field. In this review we describe the current molecular networking-based metabolite analysis methods that are widely applied or implementable in natural products discovery research, metabolomics, and related fields. The main objective of this review was to summarize strategies that can be rapidly implemented as alternative de-replication approaches for efficient natural products discovery and to list examples of successful applications that combine networking with other techniques.

Stackelberg Game Approach for Service Selection in UAV Networks.

Nowadays, mobile devices are expected to perform a growing number of tasks, whose complexity is also increasing significantly. However, despite great technological improvements in the last decade, such devices still have limitations in terms of processing power and battery lifetime. In this context, mobile edge computing (MEC) emerges as a possible solution to address such limitations, being able to provide on-demand services to the customer, and bringing closer several services published in the cloud with a reduced cost and fewer security concerns. On the other hand, Unmanned Aerial Vehicle (UAV) networking emerged as a paradigm offering flexible services, new ephemeral applications such as safety and disaster management, mobile crowd-sensing, and fast delivery, to name a few. However, to efficiently use these services, discovery and selection strategies must be taken into account. In this context, discovering the services made available by a UAV-MEC network, and selecting the best services among those available in a timely and efficient manner, can become a challenging task. To face these issues, game theory methods have been proposed in the literature that perfectly suit the case of UAV-MEC services by modeling this challenge as a Stackelberg game, and using existing approaches to find the solution for such a game aiming at an efficient services' discovery and service selection. Hence, the goal of this paper is to propose Stackelberg-game-based solutions for service discovery and selection in the context of UAV-based mobile edge computing. Simulations results conducted using the NS-3 simulator highlight the efficiency of our proposed game in terms of price and QoS metrics.

Thermoelectric Borides: Review and Future Perspectives

AbstractThe conversion of waste heat into electricity plays a vital role in energy harvesting since conventional non‐renewable energy sources have been rapidly approaching the limit of utilization. The thermoelectric (TE) technology relies on converting temperature difference (or waste heat) into electricity or vice versa. The performance of TE materials is gauged by zT = S2T/ρκ, where S, ρ, κ, and T represent the Seebeck coefficient, the electrical resistivity, the total thermal conductivity, and the absolute temperature, respectively. Boron‐based compounds are considered refractory materials due to their high melting points and advanced chemical and mechanical stability. Besides high thermal stability, many borides exhibit intrinsically low thermal conductivity. The discovery of good TE efficiency in boron carbides, where unexpectedly high Seebeck coefficients (S(T) ≈ 250 µVK−1) are observed at high charge carrier densities (≈1021 cm−3), has induced an intense curiosity for the TE properties of other boron‐based compounds. Intermetallic borides with high boron content (>65 at%) are of particular interest among all known TE materials due to their great potential in TE generators, which can operate under extreme environmental conditions (T > 1000 K). This article reviews the TE properties of borides and state‐of‐art TE materials upon boron/borides doping.

τ±νγγ and to probe the fermiophobic Higgs boson with high cutoff scales

The light fermiophobic Higgs boson hf in the type-I two-Higgs-doublet model can evade the current search programs at the LHC since its production through the quark-antiquark annihilation and gluon fusion is not feasible. The particle can be more elusive if the model retains stability up to the Planck scale because the efficient discovery channels are missing from the existing search chart. Through the comprehensive scanning, we show that all the viable parameter points with the Planck cutoff scale require {m}_{h_f}in [80, 120] GeV and {M}_{A/{H}^{pm }}in [90, 150] GeV. Since hfhf→ γγW+W− and H±→ τ±ν/hfW± are dominant in this case, two final states are more efficient to probe hf than the conventional search mode of 4γ + W±/Z. One is τ±νγγ from pp → H±(→ τ±ν)hf(→ γγ) and the other is (ℓ± = e±, μ±) from pp → H±(→ hfW±)hf→ γγW+W−W±, pp → H±(→ hfW±)A(→ hfZ) → γγW+W−W±Z, and pp → H+(→ hfW+)H−(→ hfW−) → γγW+W−W+W−. The inclusive consists of a same-sign dilepton, two prompt photons, and missing transverse energy. We perform the signal-background analysis at the detector level. With the total integrated luminosity of 300 fb−1 and the 5% background uncertainty, two proposed channels at the 14 TeV LHC yield signal significances above five in the entire viable parameter space of the fermiophobic type-I with a high cutoff scale.

Fusarium-Derived Secondary Metabolites with Antimicrobial Effects.

Fungal microbes are important in the creation of new drugs, given their unique genetic and metabolic diversity. As one of the most commonly found fungi in nature, Fusarium spp. has been well regarded as a prolific source of secondary metabolites (SMs) with diverse chemical structures and a broad spectrum of biological properties. However, little information is available concerning their derived SMs with antimicrobial effects. By extensive literature search and data analysis, as many as 185 antimicrobial natural products as SMs had been discovered from Fusarium strains by the end of 2022. This review first provides a comprehensive analysis of these substances in terms of various antimicrobial effects, including antibacterial, antifungal, antiviral, and antiparasitic. Future prospects for the efficient discovery of new bioactive SMs from Fusarium strains are also proposed.

Efficient discovery of room temperature magnetic transition metal monolayers assisted by artificial neural network

The low-dimensional materials have opened a new window to new emerging science and technological applications. Recently the discovery of new two-dimensional (2D) monolayers with long-range magnetic order by theoretical method gets more attention. The key question is how to effectively explore different possible 2D magnetic configurations. However, the computational cost of optimizing and electronically analyzing a a large number of compounds on the ab initio level is very high. A practical method is the combination of density functional theory and machine learning techniques to explore a large number of 2D monolayers. As a proof of principle, the method is implemented for looking at new high transition point 2D magnetic monolayers among transition metals dichalcogenides, dihalides, and chalcogenides/halides TMXY. Our study is intended to provide physical insights into the magnetic properties in two dimensions and introduce new room-temperature magnetic monolayers. For instance, our results show that the effective electronic screening and magnetic transition temperature in 2D monolayers are strongly dependent on the type of non-magnetic ligand atoms. Three different Neural Network (NN), Random Forest (RF), and Supporting Vector Regression (SVR) algorithms are successfully employed for quick and accurate prediction of Hubbard U, cell parameters, and the magnetic exchange parameter for a large number of TMXY (TM: V, Cr, Fe, Co, Mn, Ni, Nb and X, Y: O, S, Se, Te, Cl, Br, I) monolayers. Moreover, the binary classification is employed to determine the magnetic and atomic structure ground state. Finally, the transition temperature of TMXY monolayers is estimated from the predicted value of effective magnetic exchange parameter. The magnetic monolayers with higher transition temperatures are suggested for further theoretical and experimental investigations.

Abstract 4809: A novel microfluidic platform for PDAC organoid culture and drug screening

Abstract Pancreatic Ductal Adenocarcinoma (PDAC), the most common pancreatic cancer type, will become the second leading cause of cancer-related deaths by 2030 with mortality rates of up to 93%. Current standard-of-care for patients with PDAC includes chemotherapeutic regimens and pancreatic cancer surgery. However, only 20% of the patients are eligible for surgery due to late diagnosis. Although chemotherapeutic regimens are the leading treatment to PDAC patients, these are still very limited and the development of resistance to treatment is often observed. PDAC tumors are characterized by high-density stroma and hypovascularization, therefore these present high interstitial pressure and hypoxia. These features potentially interfere with the efficiency of chemotherapeutic drugs and highlight the urgent need for novel PDAC screening platforms. Here, we describe the establishment of PDAC organoid cultures in the MIMETAS OrganoPlate®. The aim of this project was to determine the treatment response of PDAC organoids in mono-, and co-culture with pancreatic stellate cells (PSCs) under hypoxic and normoxic conditions. To recapitulate in-vivo like conditions, the 2-lane OrganoPlate® from MIMETAS was used to study organoid growth and sensitivity to treatment. Several standard of care (chemo-)therapeutics were tested on PDAC organoids embedded within an extracellular matrix in the Organoplate® 2-lane. These were exposed to chemotherapeutic treatments for 72h. PDAC organoids showed an overall better survival when grown in co-culture with PSCs. Interestingly, organoids grown in co-culture showed a higher survival rate under hypoxic conditions. In contrast, when grown in monoculture cell viability was higher or similar under normoxic conditions than in hypoxia for the different chemotherapeutics. The OrganoPlate® 2-lane provides an excellent platform for (co-) cultivation and high-throughput phenotypic drug screening of PDAC organoids, thereby potentially enabling the development of novel in-vivo like model systems for efficient patient stratification and drug discovery. Citation Format: Marlene Geyer, Daniel Schreyer, Lisa-Marie Gaul, Désiréé Goubert, Susanne Pfeffer, Christian Pilarsky, Karla S. Queiroz. A novel microfluidic platform for PDAC organoid culture and drug screening. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4809.

Single-molecule fingerprinting of protein-drug interaction using a funneled biological nanopore

In drug discovery, efficient screening of protein-drug interactions (PDIs) is hampered by the limitations of current biophysical approaches. Here, we develop a biological nanopore sensor for single-molecule detection of proteins and PDIs using the pore-forming toxin YaxAB. Using this YaxAB nanopore, we demonstrate label-free, single-molecule detection of interactions between the anticancer Bcl-xL protein and small-molecule drugs as well as the Bak-BH3 peptide. The long funnel-shaped structure and nanofluidic characteristics of the YaxAB nanopore enable the electro-osmotic trapping of diverse folded proteins and high-resolution monitoring of PDIs. Distinctive nanopore event distributions observed in the two-dimensional (ΔI/Io-versus-IN) plot illustrate the ability of the YaxAB nanopore to discriminate individual small-molecule drugs bound to Bcl-xL from non-binders. Taken together, our results present the YaxAB nanopore as a robust platform for label-free, ultrasensitive, single-molecule detection of PDIs, opening up a possibility for low-cost, highly efficient drug discovery against diverse drug targets.

Structure-first identification of RNA elements that regulate dengue virus genome architecture and replication

The genomes of RNA viruses encode the information required for replication in host cells both in their linear sequence and in complex higher-order structures. A subset of these RNA genome structures show clear sequence conservation, and have been extensively described for well-characterized viruses. However, the extent to which viral RNA genomes contain functional structural elements-unable to be detected by sequence alone-that nonetheless are critical to viral fitness is largely unknown. Here, we devise a structure-first experimental strategy and use it to identify 22 structure-similar motifs across the coding sequences of the RNA genomes for the four dengue virus serotypes. At least 10 of these motifs modulate viral fitness, revealing a significant unnoticed extent of RNA structure-mediated regulation within viral coding sequences. These viral RNA structures promote a compact global genome architecture, interact with proteins, and regulate the viral replication cycle. These motifs are also thus constrained at the levels of both RNA structure and protein sequence and are potential resistance-refractory targets for antivirals and live-attenuated vaccines. Structure-first identification of conserved RNA structure enables efficient discovery of pervasive RNA-mediated regulation in viral genomes and, likely, other cellular RNAs.

Causal Feature Selection With Efficient Spouses Discovery

Causal feature selection has recently attracted much more attention because it can improve the interpretability of predictive models. However, the existing causal feature selection framework needs to discover the PC (i.e., parents and children) of each variable in the PC of a target variable for spouses discovery, which is time-consuming on high-dimensional data. To tackle this issue, we propose a novel <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C</u> ausal <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">F</u> eature <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">S</u> election framework with efficient spouses discovery, called CFS. Specifically, by exploiting the dependency change property between a variable and its non-PC, the proposed framework only discovers the PC of the variables in some children of the target variable for spouses discovery. Furthermore, based on the proposed CFS framework and existing PC discovery algorithms, we propose four new causal feature selection algorithms. Using benchmark Bayesian networks and real-world datasets, we experimentally validated the efficiency and accuracy of the proposed algorithms compared with seven state-of-the-art causal feature selection algorithms.

DAO to HANOI via DeSci: AI Paradigm Shifts from AlphaGo to ChatGPT

From AlphaGo to ChatGPT, the field of AI has launched a series of remarkable achievements in recent years. Analyzing, comparing, and summarizing these achievements at the paradigm level is important for future AI innovation, but has not received sufficient attention. In this paper, we give an overview and perspective on machine learning paradigms. First, we propose a paradigm taxonomy with three levels and seven dimensions from a knowledge perspective. Accordingly, we give an overview on three basic and twelve extended learning paradigms, such as Ensemble Learning, Transfer Learning, etc., with figures in unified style. We further analyze three advanced paradigms, i.e., AlphaGo, AlphaFold and ChatGPT. Second, to enable more efficient and effective scientific discovery, we propose to build a new ecosystem that drives AI paradigm shifts through the decentralized science (DeSci) movement based on decentralized autonomous organization (DAO). To this end, we design the Hanoi framework, which integrates human factors, parallel intelligence based on a combination of artificial systems and the natural world, and the DAO to inspire AI innovations.