RESUMO
In recent years, reinforcement learning (RL) has emerged as a valuable tool in drug design, offering the potential to propose and optimize molecules with desired properties. However, striking a balance between capabilities, flexibility, reliability, and efficiency remains challenging due to the complexity of advanced RL algorithms and the significant reliance on specialized code. In this work, we introduce ACEGEN, a comprehensive and streamlined toolkit tailored for generative drug design, built using TorchRL, a modern RL library that offers thoroughly tested reusable components. We validate ACEGEN by benchmarking against other published generative modeling algorithms and show comparable or improved performance. We also show examples of ACEGEN applied in multiple drug discovery case studies. ACEGEN is accessible at https://github.com/acellera/acegen-open and available for use under the MIT license.
Assuntos
Descoberta de Drogas , Descoberta de Drogas/métodos , Aprendizado de Máquina , Algoritmos , Software , Desenho de FármacosRESUMO
The accurate prediction of protein-ligand binding affinities is crucial for drug discovery. Alchemical free energy calculations have become a popular tool for this purpose. However, the accuracy and reliability of these methods can vary depending on the methodology. In this study, we evaluate the performance of a relative binding free energy protocol based on the alchemical transfer method (ATM), a novel approach based on a coordinate transformation that swaps the positions of two ligands. The results show that ATM matches the performance of more complex free energy perturbation (FEP) methods in terms of Pearson correlation but with marginally higher mean absolute errors. This study shows that the ATM method is competitive compared to more traditional methods in speed and accuracy and offers the advantage of being applicable with any potential energy function.
Assuntos
Simulação de Dinâmica Molecular , Termodinâmica , Reprodutibilidade dos Testes , Entropia , Ligação Proteica , LigantesRESUMO
Deep learning has been successfully applied to structure-based protein-ligand affinity prediction, yet the black box nature of these models raises some questions. In a previous study, we presented KDEEP, a convolutional neural network that predicted the binding affinity of a given protein-ligand complex while reaching state-of-the-art performance. However, it was unclear what this model was learning. In this work, we present a new application to visualize the contribution of each input atom to the prediction made by the convolutional neural network, aiding in the interpretability of such predictions. The results suggest that KDEEP is able to learn meaningful chemistry signals from the data, but it has also exposed the inaccuracies of the current model, serving as a guideline for further optimization of our prediction tools.
Assuntos
Redes Neurais de Computação , Proteínas , Ligantes , Proteínas/químicaRESUMO
Inflammatory bowel disease (IBD) is a group of diseases characterized by chronic and recurrent inflammation of the gastrointestinal tract. The incidence of IBD has increased significantly in past decades. The aim of this study is to review the literature on the possibility of using fecal calprotectin in the diagnosis of inflammatory bowel disease, the assessment of the severity of the disease, the prediction of a relapse and the monitoring of remission. The literature review was conducted using the PubMed and Google Scholar databases. Most of the publications included in the study are from 2013 or later. Laboratory, imaging, endoscopic and histopathological tests are used in the diagnosis of IBD. In order to confirm the diagnosis, an endoscopic examination with the collection and evaluation of histopathological specimens is required. Laboratory tests useful in the diagnosis of non-specific inflammations include non-specific tests such as the white blood cells count, ESR and CRP. Faecal calprotectin is a protein complex produced by neutrophils in the inflamed gut. The studies included in this review have shown the presence of increased levels of fecal calprotectin in patients with relapses in the course of inflammatory bowel disease. Accordingly, determination of this marker may be useful in the diagnosis of chronic abdominal pain and as screening prior to colonoscopy.
Assuntos
Fezes , Doenças Inflamatórias Intestinais , Complexo Antígeno L1 Leucocitário , Biomarcadores , Fezes/química , Humanos , Doenças Inflamatórias Intestinais/diagnóstico , Doenças Inflamatórias Intestinais/metabolismo , Complexo Antígeno L1 Leucocitário/metabolismoRESUMO
Traumatic dental injuries (TDIs) with a prevalence of 25% among school children and 33% among adults are a public health problem and can have a negative influence on the quality of life. The treatment prognosis of some teeth injuries is heavily dependent on the actions taken at the place of injury. The objective was to summarize evidence-based knowledge on the topic of TDIs and present a practical management guide for first aid in an accessible way. The authors searched the PubMed and Google Scholar databases. The review included only papers published in 2013 or later. Tooth injuries are proven to cause physical, social as well as economic consequences. The most frequent type of injury in primary dentition is avulsion, whereas crown fractures are most common in permanent dentition. TDIs occur most often at home and in school. Certain risk factors for TDIs were identified which include, among others, male gender, younger age, obesity. The general community knowledge of correct first aid in case of dental trauma is limited. Guidelines published by the International Association of Dental Traumatology include practical recommendations for first aid after avulsion. Permanent teeth should be replanted immediately at the accident site, whereas primary teeth should not be replanted when avulsed. Broken teeth fragments ought always to be collected if possible. After dental trauma it is vital that the patients seek professional help. Measures preventing TDIs (e.g., mouthguards) should be encouraged. It is of great importance that parents, teachers, guardians or bystanders witnessing a TDI are equipped to assist after a dental trauma or give advice on first aid when needed. Raising public awareness on the topic of dental injuries is a strongly advised general objective.
Assuntos
Avulsão Dentária , Fraturas dos Dentes , Traumatismos Dentários , Adulto , Criança , Dentição Permanente , Humanos , Masculino , Qualidade de Vida , Avulsão Dentária/complicações , Avulsão Dentária/terapia , Fraturas dos Dentes/etiologia , Fraturas dos Dentes/terapia , Traumatismos Dentários/epidemiologia , Traumatismos Dentários/etiologia , Traumatismos Dentários/terapiaRESUMO
Fragment-based drug discovery (FBDD) emerged as a disruptive technology and became established during the last two decades. Its rationality and low entry costs make it appealing, and the numerous examples of approved drugs discovered through FBDD validate the approach. However, FBDD still faces numerous challenges. Perhaps the most important one is the transformation of the initial fragment hits into viable leads. Fragment-to-lead (F2L) optimization is resource-intensive and is therefore limited in the possibilities that can be actively pursued. In silico strategies play an important role in F2L, as they can perform a deeper exploration of chemical space, prioritize molecules with high probabilities of being active and generate non-obvious ideas. Here we provide a critical overview of current in silico strategies in F2L optimization and highlight their remarkable impact. While very effective, most solutions are target- or fragment- specific. We propose that fully integrated in silico strategies, capable of automatically and systematically exploring the fast-growing available chemical space can have a significant impact on accelerating the release of fragment originated drugs.
Assuntos
Descoberta de DrogasRESUMO
The optimization of the Beetle readout ASIC and the performance of the software for the signal processing based on machine learning methods are presented. The Beetle readout chip was developed for the LHCb (Large Hadron Collider beauty) tracking detectors and was used in the VELO (Vertex Locator) during Run 1 and 2 of LHC data taking. The VELO, surrounding the LHC beam crossing region, was a leading part of the LHCb tracking system. The Beetle chip was used to read out the signal from silicon microstrips, integrating and amplifying it. The studies presented in this paper cover the optimization of its electronic configuration to achieve the lower power consumption footprint and the lower operational temperature of the sensors, while maintaining a good condition of the analogue response of the whole chip. The studies have shown that optimizing the operational temperature is possible and can be beneficial when the detector is highly irradiated. Even a single degree drop in silicon temperature can result in a significant reduction in the leakage current. Similar studies are being performed for the future silicon tracker, the Upstream Tracker (UT), which will start operating at LHC in 2021. It is expected that the inner part of the UT detector will suffer radiation damage similar to the most irradiated VELO sensors in Run 2. In the course of analysis we also developed a general approach for the pulse shape reconstruction using an ANN approach. This technique can be reused in case of any type of front-end readout chip.
Assuntos
Besouros , Animais , Simulação por Computador , Aprendizado de Máquina , Processamento de Sinais Assistido por Computador , SilícioRESUMO
The prediction of a ligand's binding mode into its macromolecular target is essential in structure-based drug discovery. Even though tremendous effort has been made to address this problem, most of the developed tools work similarly, trying to predict the binding free energy associated with each particular binding mode. In this study, we decided to abandon this criterion, following structural stability instead. This view, implemented in a novel computational workflow, quantifies the steepness of the local energy minimum associated with each potential binding mode. Surprisingly, the protocol outperforms docking scoring functions in case of fragments (ligands with MW < 300 Da) and is as good as docking for drug-like molecules. It also identifies substructures that act as structural anchors, predicting their binding mode with particular accuracy. The results open a new physical perspective for binding mode prediction, which can be combined with existing thermodynamic-based approaches.
Assuntos
Descoberta de Drogas , Proteínas , Sítios de Ligação , Ligantes , Ligação Proteica , Proteínas/metabolismo , TermodinâmicaRESUMO
SkeleDock is a scaffold docking algorithm which uses the structure of a protein-ligand complex as a template to model the binding mode of a chemically similar system. This algorithm was evaluated in the D3R Grand Challenge 4 pose prediction challenge, where it achieved competitive performance. Furthermore, we show that if crystallized fragments of the target ligand are available then SkeleDock can outperform rDock docking software at predicting the binding mode. This Application Note also addresses the capacity of this algorithm to model macrocycles and deal with scaffold hopping. SkeleDock can be accessed at https://playmolecule.org/SkeleDock/.
Assuntos
Desenho de Fármacos , Sítios de Ligação , Cristalografia por Raios X , Bases de Dados de Proteínas , Ligantes , Simulação de Acoplamento Molecular , Ligação Proteica , Conformação Proteica , TermodinâmicaRESUMO
Coarse graining enables the investigation of molecular dynamics for larger systems and at longer timescales than is possible at an atomic resolution. However, a coarse graining model must be formulated such that the conclusions we draw from it are consistent with the conclusions we would draw from a model at a finer level of detail. It has been proved that a force matching scheme defines a thermodynamically consistent coarse-grained model for an atomistic system in the variational limit. Wang et al. [ACS Cent. Sci. 5, 755 (2019)] demonstrated that the existence of such a variational limit enables the use of a supervised machine learning framework to generate a coarse-grained force field, which can then be used for simulation in the coarse-grained space. Their framework, however, requires the manual input of molecular features to machine learn the force field. In the present contribution, we build upon the advance of Wang et al. and introduce a hybrid architecture for the machine learning of coarse-grained force fields that learn their own features via a subnetwork that leverages continuous filter convolutions on a graph neural network architecture. We demonstrate that this framework succeeds at reproducing the thermodynamics for small biomolecular systems. Since the learned molecular representations are inherently transferable, the architecture presented here sets the stage for the development of machine-learned, coarse-grained force fields that are transferable across molecular systems.
RESUMO
Along with a growing interest in mRNA-based gene therapies, efforts are increasingly focused on reaching the full translational potential of mRNA, as a major obstacle for in vivo applications is sufficient expression of exogenously delivered mRNA. One method to overcome this limitation is chemically modifying the 7-methylguanosine cap at the 5' end of mRNA (m7Gppp-RNA). We report a novel class of cap analogs designed as reagents for mRNA modification. The analogs carry a 1,2-dithiodiphosphate moiety at various positions along a tri- or tetraphosphate bridge, and thus are termed 2S analogs. These 2S analogs have high affinities for translation initiation factor 4E, and some exhibit remarkable resistance against the SpDcp1/2 decapping complex when introduced into RNA. mRNAs capped with 2S analogs combining these two features exhibit high translation efficiency in cultured human immature dendritic cells. These properties demonstrate that 2S analogs are potentially beneficial for mRNA-based therapies such as anti-cancer immunization.
Assuntos
Difosfatos/química , Biossíntese de Proteínas , Análogos de Capuz de RNA , Capuzes de RNA , RNA Mensageiro/química , RNA Mensageiro/genética , Proteínas de Ligação a DNA/metabolismo , Células Dendríticas , Humanos , Estrutura Molecular , Ligação Proteica , Análogos de Capuz de RNA/síntese química , Fatores de Transcrição/metabolismoRESUMO
In search of new opportunities to develop Trypanosoma brucei phosphodiesterase B1 (TbrPDEB1) inhibitors that have selectivity over the off-target human PDE4 (hPDE4), different stages of a fragment-growing campaign were studied using a variety of biochemical, structural, thermodynamic, and kinetic binding assays. Remarkable differences in binding kinetics were identified and this kinetic selectivity was explored with computational methods, including molecular dynamics and interaction fingerprint analyses. These studies indicate that a key hydrogen bond between GlnQ.50 and the inhibitors is exposed to a water channel in TbrPDEB1, leading to fast unbinding. This water channel is not present in hPDE4, leading to inhibitors with a longer residence time. The computer-aided drug design protocols were applied to a recently disclosed TbrPDEB1 inhibitor with a different scaffold and our results confirm that shielding this key hydrogen bond through disruption of the water channel represents a viable design strategy to develop more selective inhibitors of TbrPDEB1. Our work shows how computational protocols can be used to understand the contribution of solvent dynamics to inhibitor binding, and our results can be applied in the design of selective inhibitors for homologous PDEs found in related parasites.
RESUMO
Developing accurate and efficient coarse-grained representations of proteins is crucial for understanding their folding, function, and interactions over extended time scales. Our methodology involves simulating proteins with molecular dynamics and utilizing the resulting trajectories to train a neural network potential through differentiable trajectory reweighting. Remarkably, this method requires only the native conformation of proteins, eliminating the need for labeled data derived from extensive simulations or memory-intensive end-to-end differentiable simulations. Once trained, the model can be employed to run parallel molecular dynamics simulations and sample folding events for proteins both within and beyond the training distribution, showcasing its extrapolation capabilities. By applying Markov state models, native-like conformations of the simulated proteins can be predicted from the coarse-grained simulations. Owing to its theoretical transferability and ability to use solely experimental static structures as training data, we anticipate that this approach will prove advantageous for developing new protein force fields and further advancing the study of protein dynamics, folding, and interactions.
Assuntos
Simulação de Dinâmica Molecular , Proteínas , Proteínas/química , Conformação Proteica , Aprendizado de Máquina , Dobramento de ProteínaRESUMO
The accurate prediction of protein-ligand binding affinities is crucial for drug discovery. Alchemical free energy calculations have become a popular tool for this purpose. However, the accuracy and reliability of these methods can vary depending on the methodology. In this study, we evaluate the performance of a relative binding free energy protocol based on the alchemical transfer method (ATM), a novel approach based on a coordinate transformation that swaps the positions of two ligands. The results show that ATM matches the performance of more complex free energy perturbation (FEP) methods in terms of Pearson correlation, but with marginally higher mean absolute errors. This study shows that the ATM method is competitive compared to more traditional methods in speed and accuracy and offers the advantage of being applicable with any potential energy function.
RESUMO
A generalized understanding of protein dynamics is an unsolved scientific problem, the solution of which is critical to the interpretation of the structure-function relationships that govern essential biological processes. Here, we approach this problem by constructing coarse-grained molecular potentials based on artificial neural networks and grounded in statistical mechanics. For training, we build a unique dataset of unbiased all-atom molecular dynamics simulations of approximately 9 ms for twelve different proteins with multiple secondary structure arrangements. The coarse-grained models are capable of accelerating the dynamics by more than three orders of magnitude while preserving the thermodynamics of the systems. Coarse-grained simulations identify relevant structural states in the ensemble with comparable energetics to the all-atom systems. Furthermore, we show that a single coarse-grained potential can integrate all twelve proteins and can capture experimental structural features of mutated proteins. These results indicate that machine learning coarse-grained potentials could provide a feasible approach to simulate and understand protein dynamics.
Assuntos
Aprendizado de Máquina , Física , Termodinâmica , Proteínas Mutadas de Ataxia Telangiectasia , Simulação de Dinâmica MolecularRESUMO
In this work, the authors analyse the influence of the order and range of sequential movements of a crane's working members on the accuracy of the final cargo positioning. The analysis was conducted on the basis of a specially developed method in which the authors proposed the introduction of a geometrical indicator of positioning the load in the intermediate positions (after completing each movement sequence) and in the target position, depending on the adopted control strategy and the accuracy of kinematic input of the working mechanisms (powered mechanisms). A mathematical model was presented to enable the accuracy of unidirectional positioning of the crane's working members when conducting sequential movements controlled through the rotation of the crane column, inner and outer boom, and retractable stages of the six-section telescopic boom. Sample results of the numerical simulations showing the influence of the assumed kinematic inputs of the crane members on the accuracy of unidirectional angular and linear positioning and, consequently, on the accuracy of the final positioning of the transported cargo, were presented. Moreover, an indicator of the cargo positioning accuracy dependent on the location of an operator or a video camera and the trajectory of the cargo was developed, allowing the formulation of application conclusions.
RESUMO
This paper presents the fundamentals of the design and applications of new worm gear drive solutions, which enable the minimisation of backlash and are characterised by higher kinematic accuracy. Different types of worm surfaces are briefly outlined. Technological problems concerning the principles of achieving a high degree of precision in machining are also described. Special attention is paid to the shaping of conical helical surfaces. Increasing the manufacturing precision of drive components allows one to achieve both lower backlash values and lower levels of its dispersion. However, this does not ensure that backlash can be eliminated, with its value being kept low during longer periods of operation. This is important in positioning systems and during recurrent operations. Various design solutions for drives in which it is possible to reduce backlash are presented. Results of experiments of a worm gear drive with a worm axially adaptive only locally, in its central section, are presented. In this solution, it is possible to reduce backlash by introducing adjustment settings without disassembling the drive. An important scientific problem concerned defining the principles of achieving a compromise between the effectiveness of reducing backlash and the required load capacity of the drive. In this paper it has been shown that in worm gear drives with a locally axially adaptive worm, as well as with a worm wheel with a deformable rim, it is possible to achieve significant reduction of backlash. In high precision drives-for example, those with an average backlash value of <15 micrometers-this can enable more than a two-fold reduction of the average backlash value and more than a three-fold decrease of the standard deviation of local backlash values.
RESUMO
Gold nanoparticles (AuNPs) decorated with biologically relevant molecules have variety of applications in optical sensing of bioanalytes. Coating AuNPs with small nucleotides produces particles with high stability in water, but functionality-compatible strategies are needed to uncover the full potential of this type of conjugates. Here, we demonstrate that lipoic acid-modified dinucleotides can be used to modify AuNPs surfaces in a controllable manner to produce conjugates that are stable in aqueous buffers and biological mixtures and capable of interacting with nucleotide-binding proteins. Using this strategy we obtained AuNPs decorated with 7-methylguanosine mRNA 5' cap analogs and showed that they bind cap-specific protein, eIF4E. AuNPs decorated with non-functional dinucleotides also interacted with eIF4E, albeit with lower affinity, suggesting that eIF4E binding to cap-decorated AuNPs is partially mediated by unspecific ionic interactions. This issue was overcome by applying lipoic-acid-Tris conjugate as a charge-neutral diluting molecule. Tris-Lipo-diluted cap-AuNPs conjugates interacted with eIF4E in fully specific manner, enabling design of functional tools. To demonstrate the potential of these conjugates in protein sensing, we designed a two-component eIF4E sensing system consisting of cap-AuNP and 4E-BP1-AuNP conjugates, wherein 4E-BP1 is a short peptide derived from 4E-BP protein that specifically binds eIF4E at a site different to that of the 5' cap. This system facilitated controlled aggregation, in which eIF4E plays the role of the agent that crosslinks two types of AuNP, thereby inducing a naked-eye visible absorbance redshift. The reported AuNPs-nucleotide conjugation method based on lipoic acid affinity for gold, can be harnessed to obtain other types of nucleotide-functionalized AuNPs, thereby paving the way to studying other nucleotide-binding proteins.
Assuntos
Proteínas Adaptadoras de Transdução de Sinal/metabolismo , Proteínas de Transporte/metabolismo , Proteínas de Ciclo Celular/metabolismo , Ouro/química , Nanopartículas Metálicas/química , Nucleotídeos/metabolismo , Fragmentos de Peptídeos/química , Capuzes de RNA/química , Proteínas Adaptadoras de Transdução de Sinal/genética , Técnicas Biossensoriais/métodos , Proteínas de Transporte/genética , Proteínas de Ciclo Celular/genética , Humanos , Nucleotídeos/química , Ligação Proteica , Biossíntese de Proteínas , Capuzes de RNA/genéticaRESUMO
Molecular dynamics simulations provide a mechanistic description of molecules by relying on empirical potentials. The quality and transferability of such potentials can be improved leveraging data-driven models derived with machine learning approaches. Here, we present TorchMD, a framework for molecular simulations with mixed classical and machine learning potentials. All force computations including bond, angle, dihedral, Lennard-Jones, and Coulomb interactions are expressed as PyTorch arrays and operations. Moreover, TorchMD enables learning and simulating neural network potentials. We validate it using standard Amber all-atom simulations, learning an ab initio potential, performing an end-to-end training, and finally learning and simulating a coarse-grained model for protein folding. We believe that TorchMD provides a useful tool set to support molecular simulations of machine learning potentials. Code and data are freely available at github.com/torchmd.
RESUMO
Computer-aided methods have been broadly used in pharmaceutical research to identify potential ligands and design effective therapeutics. Most of the approaches rely on the binding affinity prediction and approximate thermodynamic properties of the system. Our alternative approach focuses on structural stability, provided by native protein-ligand interactions, in particular hydrogen bonds. Based on this idea, we designed new fast computational method, called dynamic undocking (DUck), that evaluates stability by calculating the work necessary to break the most important native contact in a ligand-receptor complex. This property is effective in distinguishing true ligands from decoys and is orthogonal to currently existing docking methods, thus making it exceptionally useful in virtual screening. Here, we present a protocol suitable for DUck's application in drug design strategy, as well as notes that will help to solve common problems addressed by users.