Biomimetic molecular design tools that learn, evolve, and adapt.
Ontology highlight
ABSTRACT:
SUBMITTER: Winkler DA
PROVIDER: S-EPMC5496546 | biostudies-other | 2017
REPOSITORIES: biostudies-other
Similar Datasets
Project description:We present four experiments investigating adaptation to a regional grammatical structure through reading exposure, using both the needs + past participle construction (e.g., The car needs washed) and the double modal construction (e.g. You might could go there). In each experiment, participants read two stories containing informal dialogue. Half of the participants were exposed to one of the regional constructions and half were not. Those readers exposed to the regional constructions adapted, gradually reading the novel constructions faster over 9 to 15 exemplars. The degree to which the exposed group learned the construction was tested in two ways. In the first two experiments, learning was measured by comparing reading times to acceptable and unacceptable variants of the novel constructions. Readers did not learn either the verb tense rule for the needs construction (Experiment 1) or a simple ordering rule for double modal constructions (Experiment 2). Similarly, in Experiments 3 and 4, metalinguistic judgments used to test learning revealed that participants had failed to acquire the regional grammar of either novel construction. These experiments suggest that the adaptation effects reflect learning some general properties of the experimental stimuli, not learning the syntactic constructions themselves.
| S-EPMC10081778 | biostudies-literature
Project description:Biomaterials such as self-assembling biological complexes have demonstrated a variety of applications in materials science and nanotechnology. The functionality of protein-based materials, however, is often limited by the absence or locations of specific chemical conjugation sites. In this investigation, we developed a new strategy for loading organic molecules into the hollow cavity of a protein nanoparticle that relies only on non-covalent interactions, and we demonstrated its applicability in drug delivery. Based on a biomimetic model that incorporates multiple phenylalanines to create a generalized binding site, we retained and delivered the antitumor compound doxorubicin by redesigning a caged protein scaffold. Through an iterative combination of structural modeling and protein engineering, we obtained new variants of the E2 subunit of pyruvate dehydrogenase with varying levels of drug-carrying capabilities. We found that an increasing number of introduced phenylalanines within the scaffold cavity generally resulted in greater drug loading capacities. Drug loading levels could be achieved that were greater than conventional nanoparticle delivery systems. These protein nanoparticles containing doxorubicin were taken up by breast cancer cells and induced significant cell death. Our novel approach demonstrates a universal strategy to design de novo hydrophobic binding domains within protein-based scaffolds for molecular encapsulation and transport, and it broadens the ability to attach guest molecules to this class of materials.
| S-EPMC3603581 | biostudies-literature
Project description:Recent years have seen a tremendous progress in the elucidation of experimental structural information for G-protein coupled receptors (GPCRs). Although for the vast majority of pharmaceutically relevant GPCRs structural information is still accessible only by homology models the steadily increasing amount of structural information fosters the application of structure-based drug design tools for this important class of drug targets. In this article we focus on the application of molecular dynamics (MD) simulations in GPCR drug discovery programs. Typical application scenarios of MD simulations and their scope and limitations will be described on the basis of two selected case studies, namely the binding of small molecule antagonists to the human CC chemokine receptor 3 (CCR3) and a detailed investigation of the interplay between receptor dynamics and solvation for the binding of small molecules to the human muscarinic acetylcholine receptor 3 (hM3R).
| S-EPMC4334948 | biostudies-literature
Project description:A long-standing goal in artificial intelligence is creating agents that can learn a variety of different skills for different problems. In the artificial intelligence subfield of neural networks, a barrier to that goal is that when agents learn a new skill they typically do so by losing previously acquired skills, a problem called catastrophic forgetting. That occurs because, to learn the new task, neural learning algorithms change connections that encode previously acquired skills. How networks are organized critically affects their learning dynamics. In this paper, we test whether catastrophic forgetting can be reduced by evolving modular neural networks. Modularity intuitively should reduce learning interference between tasks by separating functionality into physically distinct modules in which learning can be selectively turned on or off. Modularity can further improve learning by having a reinforcement learning module separate from sensory processing modules, allowing learning to happen only in response to a positive or negative reward. In this paper, learning takes place via neuromodulation, which allows agents to selectively change the rate of learning for each neural connection based on environmental stimuli (e.g. to alter learning in specific locations based on the task at hand). To produce modularity, we evolve neural networks with a cost for neural connections. We show that this connection cost technique causes modularity, confirming a previous result, and that such sparsely connected, modular networks have higher overall performance because they learn new skills faster while retaining old skills more and because they have a separate reinforcement learning module. Our results suggest (1) that encouraging modularity in neural networks may help us overcome the long-standing barrier of networks that cannot learn new skills without forgetting old ones, and (2) that one benefit of the modularity ubiquitous in the brains of natural animals might be to alleviate the problem of catastrophic forgetting.
| S-EPMC4383335 | biostudies-literature
Project description:The luminescent metal-organic complexes of rare earth metals are advanced materials with wide application potential in chemistry, biology, and medicine. The luminescence of these materials is due to a rare photophysical phenomenon called antenna effect, in which the excited ligand transmits its energy to the emitting levels of the metal. However, despite the attractive photophysical properties and the intriguing from a fundamental point of view antenna effect, the theoretical molecular design of new luminescent metal-organic complexes of rare earth metals is relatively limited. Our computational study aims to contribute in this direction, and we model the excited state properties of four new phenanthroline-based complexes of Eu(III) using the TD-DFT/TDA approach. The general formula of the complexes is EuL2A3, where L is a phenanthroline with -2-CH3O-C6H4, -2-HO-C6H4, -C6H5 or -O-C6H5 substituent at position 2 and A is Cl- or NO3-. The antenna effect in all newly proposed complexes is estimated as viable and is expected to possess luminescent properties. The relationship between the electronic properties of the isolated ligands and the luminescent properties of the complexes is explored in detail. Qualitative and quantitative models are derived to interpret the ligand-to-complex relation, and the results are benchmarked with respect to available experimental data. Based on the derived model and common molecular design criteria for efficient antenna ligands, we choose phenanthroline with -O-C6H5 substituent to perform complexation with Eu(III) in the presence of NO3¯. Experimental results for the newly synthesized Eu(III) complex are reported with a luminescent quantum yield of about 24% in acetonitrile. The study demonstrates the potential of low-cost computational models for discovering metal-organic luminescent materials.
| S-EPMC10221419 | biostudies-literature
Project description:Effective biologic therapeutics require binding affinities that are fine-tuned to their disease-related molecular target. The ADAPT (Assisted Design of Antibody and Protein Therapeutics) platform aids in the selection of mutants that improve/modulate the affinity of antibodies and other biologics. It uses a consensus z-score from three scoring functions and interleaves computational predictions with experimental validation, significantly enhancing the robustness of the design and selection of mutants. The platform was tested on three antibody Fab-antigen systems that spanned a wide range of initial binding affinities: bH1-VEGF-A (44 nM), bH1-HER2 (3.6 nM) and Herceptin-HER2 (0.058 nM). Novel triple mutants were obtained that exhibited 104-, 46- and 32-fold improvements in binding affinity for each system, respectively. Moreover, for all three antibody-antigen systems over 90% of all the intermediate single and double mutants that were designed and tested showed higher affinities than the parent sequence. The contributions of the individual mutants to the change in binding affinity appear to be roughly additive when combined to form double and triple mutants. The new interactions introduced by the affinity-enhancing mutants included long-range electrostatics as well as short-range nonpolar interactions. This diversity in the types of new interactions formed by the mutants was reflected in SPR kinetics that showed that the enhancements in affinities arose from increasing on-rates, decreasing off-rates or a combination of the two effects, depending on the mutation. ADAPT is a very focused search of sequence space and required only 20-30 mutants for each system to be made and tested to achieve the affinity enhancements mentioned above.
| S-EPMC5531539 | biostudies-literature
Project description:BackgroundThe combination of experimental evolution with whole-genome resequencing of pooled individuals, also called evolve and resequence (E&R) is a powerful approach to study the selection processes and to infer the architecture of adaptive variation. Given the large potential of this method, a range of software tools were developed to identify selected SNPs and to measure their selection coefficients.ResultsIn this benchmarking study, we compare 15 test statistics implemented in 10 software tools using three different scenarios. We demonstrate that the power of the methods differs among the scenarios, but some consistently outperform others. LRT-1, CLEAR, and the CMH test perform best despite LRT-1 and the CMH test not requiring time series data. CLEAR provides the most accurate estimates of selection coefficients.ConclusionThis benchmark study will not only facilitate the analysis of already existing data, but also affect the design of future data collections.
| S-EPMC6694636 | biostudies-literature
Project description:Standing genetic variation provides a rich reservoir of potentially useful mutations facilitating the adaptation to novel environments. Experimental evolution studies have demonstrated that rapid and strong phenotypic responses to selection can also be obtained in the laboratory. When combined with the next-generation sequencing technology, these experiments promise to identify the individual loci contributing to adaption. Nevertheless, until now, very little is known about the design of such evolve & resequencing (E&R) studies. Here, we use forward simulations of entire genomes to evaluate different experimental designs that aim to maximize the power to detect selected variants. We show that low linkage disequilibrium in the starting population, population size, duration of the experiment, and the number of replicates are the key factors in determining the power and accuracy of E&R studies. Furthermore, replication of E&R is more important for detecting the targets of selection than increasing the population size. Using an optimized design, beneficial loci with a selective advantage as low as s = 0.005 can be identified at the nucleotide level. Even when a large number of loci are selected simultaneously, up to 56% can be reliably detected without incurring large numbers of false positives. Our computer simulations suggest that, with an adequate experimental design, E&R studies are a powerful tool to identify adaptive mutations from standing genetic variation and thereby provide an excellent means to analyze the trajectories of selected alleles in evolving populations.
| S-EPMC3907048 | biostudies-literature
Project description:Deep generative models of molecules have grown immensely in popularity, trained on relevant datasets, these models are used to search through chemical space. The downstream utility of generative models for the inverse design of novel functional compounds, depends on their ability to learn a training distribution of molecules. The most simple example is a language model that takes the form of a recurrent neural network and generates molecules using a string representation. Since their initial use, subsequent work has shown that language models are very capable, in particular, recent research has demonstrated their utility in the low data regime. In this work, we investigate the capacity of simple language models to learn more complex distributions of molecules. For this purpose, we introduce several challenging generative modeling tasks by compiling larger, more complex distributions of molecules and we evaluate the ability of language models on each task. The results demonstrate that language models are powerful generative models, capable of adeptly learning complex molecular distributions. Language models can accurately generate: distributions of the highest scoring penalized LogP molecules in ZINC15, multi-modal molecular distributions as well as the largest molecules in PubChem. The results highlight the limitations of some of the most popular and recent graph generative models- many of which cannot scale to these molecular distributions.
| S-EPMC9174447 | biostudies-literature
Project description:Many species of bat emit acoustic signals and use information carried by echoes reflecting from nearby objects to navigate and forage. It is widely documented that echolocating bats adjust the features of sonar calls in response to echo feedback; however, it remains unknown whether audiovocal feedback contributes to sonar call design. Audiovocal feedback refers to the monitoring of one's own vocalizations during call production and has been intensively studied in nonecholocating animals. Audiovocal feedback not only is a necessary component of vocal learning but also guides the control of the spectro-temporal structure of vocalizations. Here, we show that audiovocal feedback is directly involved in the echolocating bat's control of sonar call features. As big brown bats tracked targets from a stationary position, we played acoustic jamming signals, simulating calls of another bat, timed to selectively perturb audiovocal feedback or echo feedback. We found that the bats exhibited the largest call-frequency adjustments when the jamming signals occurred during vocal production. By contrast, bats did not show sonar call-frequency adjustments when the jamming signals coincided with the arrival of target echoes. Furthermore, bats rapidly adapted sonar call design in the first vocalization following the jamming signal, revealing a response latency in the range of 66 to 94 ms. Thus, bats, like songbirds and humans, rely on audiovocal feedback to structure sonar signal design.
| S-EPMC5642720 | biostudies-literature