15,635 research outputs found

    LS-MolGen: Ligand-and-Structure Dual-Driven Deep Reinforcement Learning for Target-Specific Molecular Generation Improves Binding Affinity and Novelty

    No full text
    Deep learning-based molecular generative models have garnered considerable interest in the field of de novo drug design. However, most extant models focus on either ligand-based or structure-based strategies, thereby failing to effectively harness the combined knowledge derived from both ligands and the structure of the binding target. In this article, we introduce LS-MolGen, a novel ligand and structure-integrated molecular generative model. This model synergistically combines representation learning, transfer learning, and reinforcement learning. The targeted knowledge assimilation from transfer learning, coupled with an advanced exploration strategy in reinforcement learning, empowers LS-MolGen to efficiently generate novel and high-affinity molecules efficiently. The comparable performance of our model is affirmed through multiple evaluations, including EGFR, DRD3, CDK2, AA2AR, ADRB2, and a dedicated case study of inhibitor design for SARS-CoV-2 Mpro. The results indicate that LS-MolGen performs better than other ligand-based or structure-based generative models in de novo designing promising compounds with novel scaffolds and high binding affinity. This proof-of-concept study signifies the potential of our ligand- and structure-based generative model, LS-MolGen, as a promising new tool for target-specific molecular generation and drug design

    Oligostilbenes from Vitis heyneana

    No full text
    A novel tetrastilbene, heyneanol A, was isolated from stems of Vitis heyneana together with the previously known ampelopsin C, ampelopsin A and (+)-epsilon-viniferin. Its complex polycyclic structure was elucidated by 1D and 2D NMR analyses

    Auditory fear memory retrieval requires BLA-LS and LS-VMH circuitries via GABAergic and dopaminergic neurons

    No full text
    Abstract Fear and associated learning and memory are critical for developing defensive behavior. Excessive fear and anxiety are important components of post-traumatic stress disorder. However, the neurobiology of fear conditioning, especially tone-related fear memory retrieval, has not been clearly defined, which limits specific intervention development for patients with excessive fear and anxiety. Here, we show that auditory fear memory retrieval stimuli activate multiple brain regions including the lateral septum (LS). Inhibition of the LS and the connection between basolateral amygdala (BLA) and LS or between LS and ventromedial nucleus of the hypothalamus (VMH) attenuates tone-related fear conditioning and memory retrieval. Inhibiting GABAergic neurons or dopaminergic neurons in the LS also attenuates tone-related fear conditioning. Our data further show that fear conditioning is inhibited by blocking orexin B signaling in the LS. Our results indicate that the neural circuitries BLA–LS and LS-VMH are critical for tone-related fear conditioning and memory retrieval, and that GABAergic neurons, dopaminergic neurons and orexin signaling in the LS participate in this auditory fear conditioning

    Movie S1_Optical layout of LS-FIS.

    No full text
    Optical layout of LS-FIS

    Movie S2_Imaging procedure of LS-FIS.

    No full text
    Imaging procedure of LS-FIS

    Author Attributions in Medieval Text Collections: An Exploration

    No full text
    This article examines the role and function of author attributions in multi-text manuscripts containing Dutch, English, French or German short verse narratives. The findings represent one strand of the investigations undertaken by the cross-European project ‘The Dynamics of the Medieval Manuscript’, which analysed the dissemination of short verse narratives and the principles of organisation underlying the compilation of text collections. Whilst short verse narratives are more commonly disseminated anonymously, there are manuscripts in which authorship is repeatedly attributed to a text or corpus. Through six case studies, this article explores medieval concepts of authorship and how they relate to constructions of authority, whether regarding an empirical figure or a literary construction. In addition, it looks at how authorship plays a role in manuscript compilation, and at the effects of attributions (by author and/or compiler) on reception. The case studies include manuscripts from the thirteenth to fifteenth centuries, produced in a range of social and cultural contexts, and featuring some of the most important European authors of short verse narratives: Rutebeuf, Baudouin de Condé, Der Striker, Konrad von Würzberg, Willem of Hildegaersberch, and Geoffrey Chaucer. The preliminary findings contribute to our understanding of author attributions in text collections from across northern Europe and point towards future lines of enquiry into the role of authorship in medieval textual dissemination

    Measured vs. predicted values of calibration and prediction by CARS-LS-SVM and SPA-LS-SVM models, respectively.

    No full text
    (a): CARS-LS-SVM-ΔL*; (b): CARS-LS-SVM-Δa*; (c): CARS-LS-SVM-Δb*; (d): SPA-LS-SVM-ΔL*; e): SPA-LS-SVM-Δa*; (f): SPA-LS-SVM-Δb*.</p

    The clitic li at the syntax-phonology interface: An experimentally informed account.

    No full text
    Among Serbo-Croatian (SC) clitics, the interrogative li seems to be unique in showing a peculiar type of sensitivity with respect to the type of host it cliticizes (on)to. Whereas li is unproblematically hosted by a finite verb, when hosted by an l-participle, the outputs are treated as fully ungrammatical. Be it in syntax or in phonology, current accounts in the literature overgenerate by predicting the ungrammatical constructions to be grammatical, and undergenerate by predicting grammatical outputs in SC to be ungrammatical. Moreover, native speakers’ judgements regarding the l-participle li construction suggest a more gradient judgment than standardly assumed in the literature. We present results of a large-scale experiment undertaken not only because of the unclarities regarding the acceptability judgments we encountered for l-participle li, but, first and foremost, because of the lack of prior empirical investigations or established baseline of acceptability in the domain that has, otherwise, been a prolific and important domain for theorizing in SC. On the theoretical side, we argue that the crux of the solution is not due to the defectiveness of li (as previously argued), but due to the [-finite] nature of participles. Testing our hypothesis against the data obtained in a systematic manner, we further demonstrate that the gradedness in (un)grammaticality judgments are accounted for by analyzing the interplay as well as the tensions that arise at the syntax-phonology interface
    corecore