small chemical
Recently Published Documents


TOTAL DOCUMENTS

201
(FIVE YEARS 64)

H-INDEX

19
(FIVE YEARS 3)

2022 ◽  
Author(s):  
Mingjian Wen ◽  
Samuel M. Blau ◽  
Xiaowei Xie ◽  
Shyam Dwaraknath ◽  
Kristin A. Persson

Machine learning (ML) methods have great potential to transform chemical discovery by accelerating the exploration of chemical space and drawing scientific insights from data. However, modern chemical reaction ML models, such as those based on graph neural networks (GNNs), must be trained on a large amount of labelled data in order to avoid overfitting the data and thus possessing low accuracy and transferability. In this work, we propose a strategy to leverage unlabelled data to learn accurate ML models for small labelled chemical reaction data. We focus on an old and prominent problem—classifying reactions into distinct families—and build a GNN model for this task. We first pretrain the model on unlabelled reaction data using unsupervised contrastive learning and then fine-tune it on a small number of labelled reactions. The contrastive pretraining learns by making the representations of two augmented versions of a reaction similar to each other but distinct from other reactions. We propose chemically consistent reaction augmentation methods that protect the reaction center and find they are the key for the model to extract relevant information from unlabelled data to aid the reaction classification task. The transfer learned model outperforms a supervised model trained from scratch by a large margin. Further, it consistently performs better than models based on traditional rule-driven reaction fingerprints, which have long been the default choice for small datasets. In addition to reaction classification, the effectiveness of the strategy is tested on regression datasets; the learned GNN-based reaction fingerprints can also be used to navigate the chemical reaction space, which we demonstrate by querying for similar reactions. The strategy can be readily applied to other predictive reaction problems to uncover the power of unlabelled data for learning better models with a limited supply of labels.


2022 ◽  
Author(s):  
Mingjian Wen ◽  
Samuel M. Blau ◽  
Xiaowei Xie ◽  
Shyam Dwaraknath ◽  
Kristin Persson

Machine learning (ML) methods have great potential to transform chemical discovery by accelerating the exploration of chemical space and drawing scientific insights from data. However, modern chemical reaction ML models,...


2021 ◽  
Author(s):  
Jenna N Meanor ◽  
Albert J Keung ◽  
Balaji M Rao

Histone post-translational modifications are small chemical changes to histone protein structure that have cascading effects on diverse cellular functions. Detecting histone modifications and characterizing their binding partners are critical steps in understanding chromatin biochemistry and have been accessed using common reagents such as antibodies, recombinant assays, and FRET based systems. High throughput platforms could accelerate work in this field, and also could be used to engineer de novo histone affinity reagents; yet published studies on their use with histones have been noticeably sparse. Here we describe specific experimental conditions that affect binding specificities of post-translationally modified histones in classic protein engineering platforms and likely explain the relative difficulty with histone targets in these platforms. We also show that manipulating avidity of binding interactions may improve specificity of binding.


Chemosensors ◽  
2021 ◽  
Vol 9 (12) ◽  
pp. 347
Author(s):  
Tatjana Kulikova ◽  
Pavel Padnya ◽  
Igor Shiabiev ◽  
Alexey Rogov ◽  
Ivan Stoikov ◽  
...  

In this work, we investigated aggregation of native DNA and thiacalix[4]arene derivative bearing eight terminal amino groups in cone configuration using various redox probes on the glassy carbon electrode. It was shown that sorption transfer of the aggregates on the surface of the electrode covered with carbon black resulted in changes in electrostatic interactions and diffusional permeability of the surface layer. Such changes alter the signals of ferricyanide ion, methylene green and hydroquinone as redox probes to a degree depending on their specific interactions with DNA and own charge. Inclusion of DNA in the surface layer was independently confirmed by scanning electron microscopy, electrochemical impedance spectroscopy and experiments with doxorubicin as a model intercalator. Thermal denaturing of DNA affected the charge separation on the electrode interface and the signals of redox probes. Using hydroquinone, less sensitive to electrostatic interactions, made it possible to determine from 10 pM to 1.0 nM doxorubicin (limit of detection 3 pM) after 10 min incubation. Stabilizers present in the commercial medications did not alter the signal. The DNA sensors developed can find future application in the assessment of the complexes formed by DNA and macrocycles as delivery agents for small chemical species.


Author(s):  
Suyeong Han ◽  
Yongwon Jung

Nature uses a wide range of well-defined biomolecular assemblies in diverse cellular processes, where proteins are major building blocks for these supramolecular assemblies. Inspired by their natural counterparts, artificial protein-based assemblies have attracted strong interest as new bio-nanostructures, and strategies to construct ordered protein assemblies have been rapidly expanding. In this review, we provide an overview of very recent studies in the field of artificial protein assemblies, with the particular aim of introducing major assembly methods and unique features of these assemblies. Computational de novo designs were used to build various assemblies with artificial protein building blocks, which are unrelated to natural proteins. Small chemical ligands and metal ions have also been extensively used for strong and bio-orthogonal protein linking. Here, in addition to protein assemblies with well-defined sizes, protein oligomeric and array structures with rather undefined sizes (but with definite repeat protein assembly units) also will be discussed in the context of well-defined protein nanostructures. Lastly, we will introduce multiple examples showing how protein assemblies can be effectively used in various fields such as therapeutics and vaccine development. We believe that structures and functions of artificial protein assemblies will be continuously evolved, particularly according to specific application goals.


2021 ◽  
Author(s):  
Mingjian Wen ◽  
Samuel M. Blau ◽  
Xiaowei Xie ◽  
Shyam Dwaraknath ◽  
Kristin A. Persson

Machine learning (ML) methods have great potential to transform chemical discovery by accelerating the exploration of chemical space and drawing scientific insights from data. However, modern chemical reaction ML models, such as those based on graph neural networks (GNNs), must be trained on a large amount of labelled data in order to avoid overfitting the data and thus possessing low accuracy and transferability. In this work, we propose a strategy to leverage unlabelled data to learn accurate ML models for small labelled chemical reaction data. We focus on an old and prominent problem—classifying reactions into distinct families—and build a GNN model for this task. We first pretrain the model on unlabelled reaction data using unsupervised contrastive learning and then fine-tune it on a small number of labelled reactions. The contrastive pretraining learns by making the representations of two augmented versions of a reaction similar to each other but distinct from other reactions. We propose chemically consistent reaction augmentation methods that protect the reaction center and find they are the key for the model to extract relevant information from unlabelled data to aid the reaction classification task. The transfer learned model outperforms a supervised model trained from scratch by a large margin. Further, it consistently performs better than models based on traditional rule-driven reaction fingerprints, which have long been the default choice for small datasets. In addition to reaction classification, the learned GNN-based reaction fingerprints can also be used to navigate the chemical reaction space, which we demonstrate by querying for similar reactions. The strategy can be readily applied to other predictive reaction problems to uncover the power of unlabelled data for learning better models with a limited supply of labels.


Crystals ◽  
2021 ◽  
Vol 11 (11) ◽  
pp. 1397
Author(s):  
Torvid Feiler ◽  
Adam A. L. Michalchuk ◽  
Vincent Schröder ◽  
Emil List-Kratochvil ◽  
Franziska Emmerling ◽  
...  

Organic single crystals that combine mechanical flexibility and optical properties are important for developing flexible optical devices, but examples of such crystals remain scarce. Both mechanical flexibility and optical activity depend on the underlying crystal packing and the nature of the intermolecular interactions present in the solid state. Hence, both properties can be expected to be tunable by small chemical modifications to the organic molecule. By incorporating a chlorine atom, a reportedly mechanically flexible crystal of (E)-1-(4-bromo-phenyl)iminomethyl-2-hydroxyl-naphthalene (BPIN) produces (E)-1-(4-bromo-2-chloro-phenyl)iminomethyl-2-hydroxyl-naphthalene (BCPIN). BCPIN crystals show elastic bending similar to BPIN upon mechanical stress, but exhibit a remarkable difference in their optical properties as a result of the chemical modification to the backbone of the organic molecule. This work thus demonstrates that the optical properties and mechanical flexibility of molecular materials can, in principle, be tuned independently.


2021 ◽  
Vol 2 (2) ◽  
pp. 795-813
Author(s):  
Davy Sinnaeve ◽  
Abir Ben Bouzayene ◽  
Emile Ottoy ◽  
Gert-Jan Hofman ◽  
Eva Erdmann ◽  
...  

Abstract. Proline homopolymer motifs are found in many proteins; their peculiar conformational and dynamic properties are often directly involved in those proteins' functions. However, the dynamics of proline homopolymers is hard to study by NMR due to a lack of amide protons and small chemical shift dispersion. Exploiting the spectroscopic properties of fluorinated prolines opens interesting perspectives to address these issues. Fluorinated prolines are already widely used in protein structure engineering – they introduce conformational and dynamical biases – but their use as 19F NMR reporters of proline conformation has not yet been explored. In this work, we look at model peptides where Cγ-fluorinated prolines with opposite configurations of the chiral Cγ centre have been introduced at two positions in distinct polyproline segments. By looking at the effects of swapping these (4R)-fluoroproline and (4S)-fluoroproline within the polyproline segments, we were able to separate the intrinsic conformational properties of the polyproline sequence from the conformational alterations instilled by fluorination. We assess the fluoroproline 19F relaxation properties, and we exploit the latter in elucidating binding kinetics to the SH3 (Src homology 3) domain.


Author(s):  
Z. U. Zufarova ◽  
S. SH. Tashpulatov ◽  
I. V. Cherunova ◽  
Zhen Yaven

In this article, an analytical method is carried out in innovative developments in the field of diving suits with high thermal conductivity. Examples are given of the developments of the athlete and scientist Jack O’Neill, as well as the innovative developments of American chemists under the leadership of Jacopo Buongiorno from the Massachusetts Institute of Technology. Developers, together with chemists, on the basis of research, introduce small chemical elements into the pores of neoprene fabric, thus achieving the necessary thermal stability, flexibility and strength. During the analytical method, not only neprene wetsuits were studied, but also the latest developments in the field of eco-suits. This foam is made from limestone, as it is a natural material, which allows these suits to reduce the negative impact on the environment.


2021 ◽  
Author(s):  
Leehyeon Kim ◽  
Byung-Gil Lee ◽  
Min Kyung Kim ◽  
Do Hoon Kwon ◽  
Hyunmin Kim ◽  
...  

The ClpP serine peptidase is a tetradecameric degradation machine involved in many physiological processes. It becomes a competent ATP-dependent protease with Clp-ATPases. Small chemical compounds, acyldepsipeptides (ADEPs), are known to cause dysregulation and activation of ClpP without ATPases, and have potential as novel antibiotics. Previously, structural studies of ClpP from various species revealed the structural details, conformational changes, and activation mechanism. Although product release by the side exit pores has been proposed, the detailed driving force for product release remains elusive. Here, we report crystal structures of ClpP from Bacillus subtilis (BsClpP) in unforeseen ADEP-bound states. Cryo-electron microscopy structures revealed various conformational states at different pH conditions. To understand the conformational change for product release, we investigated the relationship between substrate hydrolysis and the pH lowering process. Our data, together with previous findings, provide insight into the molecular mechanism of product release by ClpP self-compartmentalizing protease.


Sign in / Sign up

Export Citation Format

Share Document