The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in 2019 has triggered an ongoing global pandemic of the severe pneumonia-like disease coronavirus disease 2019 (COVID-19)
.... The development of a vaccine is likely to take at least 12-18 months, and the typical timeline for approval of a new antiviral therapeutic agent can exceed 10 years. Thus, repurposing of known drugs could substantially accelerate the deployment of new therapies for COVID-19. Here we profiled a library of drugs encompassing approximately 12,000 clinical-stage or Food and Drug Administration (FDA)-approved small molecules to identify candidate therapeutic drugs for COVID-19. We report the identification of 100 molecules that inhibit viral replication of SARS-CoV-2, including 21 drugs that exhibit dose-response relationships. Of these, thirteen were found to harbour effective concentrations commensurate with probable achievable therapeutic doses in patients, including the PIKfyve kinase inhibitor apilimod
and the cysteine protease inhibitors MDL-28170, Z LVG CHN2, VBY-825 and ONO 5334. Notably, MDL-28170, ONO 5334 and apilimod were found to antagonize viral replication in human pneumocyte-like cells derived from induced pluripotent stem cells, and apilimod also demonstrated antiviral efficacy in a primary human lung explant model. Since most of the molecules identified in this study have already advanced into the clinic, their known pharmacological and human safety profiles will enable accelerated preclinical and clinical evaluation of these drugs for the treatment of COVID-19.
Inferring molecular structure from Nuclear Magnetic Resonance (NMR) measurements requires an accurate forward model that can predict chemical shifts from 3D structure. Current forward models are ...limited to specific molecules like proteins and state-of-the-art models are not differentiable. Thus they cannot be used with gradient methods like biased molecular dynamics. Here we use graph neural networks (GNNs) for NMR chemical shift prediction. Our GNN can model chemical shifts accurately and capture important phenomena like hydrogen bonding induced downfield shift between multiple proteins, secondary structure effects, and predict shifts of organic molecules. Previous empirical NMR models of protein NMR have relied on careful feature engineering with domain expertise. These GNNs are trained from data alone with no feature engineering yet are as accurate and can work on arbitrary molecular structures. The models are also efficient, able to compute one million chemical shifts in about 5 seconds. This work enables a new category of NMR models that have multiple interacting types of macromolecules.
This model can predict chemical shifts on proteins and small molecules purely from atom elements and coordinates. It can capture important phenomena like hydrogen bonding induced downfield shift, thus can be used to infer intermolecular interactions.
An outstanding challenge in deep learning in chemistry is its lack of interpretability. The inability of explaining why a neural network makes a prediction is a major barrier to deployment of AI ...models. This not only dissuades chemists from using deep learning predictions, but also has led to neural networks learning spurious correlations that are difficult to notice. Counterfactuals are a category of explanations that provide a rationale behind a model prediction with satisfying properties like providing chemical structure insights. Yet, counterfactuals have been previously limited to specific model architectures or required reinforcement learning as a separate process. In this work, we show a universal model-agnostic approach that can explain any black-box model prediction. We demonstrate this method on random forest models, sequence models, and graph neural networks in both classification and regression.
Generating model agnostic molecular counterfactual explanations to explain model predictions.
We present three deep learning sequence-based prediction models for peptide properties including hemolysis, solubility, and resistance to nonspecific interactions that achieve comparable results to ...the state-of-the-art models. Our sequence-based solubility predictor, MahLooL, outperforms the current state-of-the-art methods for short peptides. These models are implemented as a static website without the use of a dedicated server or cloud computing. Web-based models like this allow for accessible and effective reproducibility. Most existing approaches rely on third-party servers that typically require upkeep and maintenance. Our predictive models do not require servers, require no installation of dependencies, and work across a range of devices. The specific architecture is bidirectional recurrent neural networks. This serverless approach is a demonstration of edge machine learning that removes the dependence on cloud providers. The code and models are accessible at https://github.com/ur-whitelab/peptide-dashboard.
Large language models such as GPT-4 have been approaching human-level ability across many expert domains. GPT-4 can accomplish complex tasks in chemistry purely from English instructions, which may ...transform the future of chemistry.
Often the development of novel functional peptides is not amenable to high throughput or purely computational screening methods. Peptides must be synthesized one at a time in a process that does not ...generate large amounts of data. One way this method can be improved is by ensuring that each experiment provides the best improvement in both peptide properties and predictive modeling accuracy. Here, we study the effectiveness of active learning, optimizing experiment order, and meta-learning, transferring knowledge between contexts, to reduce the number of experiments necessary to build a predictive model. We present a multitask benchmark database of peptides designed to advance these methods for experimental design. Each task is a binary classification of peptides represented as a sequence string. We find neither active learning method tested to be better than random choice. The meta-learning method Reptile was found to improve the average accuracy across data sets. Combining meta-learning with active learning offers inconsistent benefits.
In this work, we report a study on the differential hydration of carboxybetaine and sulfobetaine using molecular simulations. The coordination number, spatial distribution, dipole orientation ...distribution, and residence time of water molecules around the positively charged group (N(CH3)3 +) and negatively charged group (COO− for carboxybetaine and SO3 − for sulfobetaine) were investigated to compare the hydration of these two betaines. The results show that the negatively charged group of sulfobetaine has more water molecules around it than that of carboxybetaine, while the water molecules around the negatively charged group of the carboxybetaine have a sharper spatial distribution, more preferential dipole orientation, and longer residence time. The behavior of water molecules around the positively charged group of sulfobetaine is similar to those around the positively charged group of carboxybetaine. For both sulfobetaine and carboxybetaine, the positively charged groups are surrounded by more water molecules than the negatively charged groups, whereas the water molecules around the negatively charged groups are more ordered than those around the positively charged ones. We also investigated the hydration free energy of these two molecules with the free energy perturbation method and found that their values are all considerably lower than that of oligo(ethylene glycol).
We characterised the clinical course, treatment and outcomes in 59 patients with relapsing myelin oligodendrocyte glycoprotein (MOG) antibody-associated demyelination.
We evaluated clinical ...phenotypes, annualised relapse rates (ARR) prior and on immunotherapy and Expanded Disability Status Scale (EDSS), in 218 demyelinating episodes from 33 paediatric and 26 adult patients.
The most common initial presentation in the cohort was optic neuritis (ON) in 54% (bilateral (BON) 32%, unilateral (UON) 22%), followed by acute disseminated encephalomyelitis (ADEM) (20%), which occurred exclusively in children. ON was the dominant phenotype (UON 35%, BON 19%) of all clinical episodes. 109/226 (48%) MRIs had no brain lesions. Patients were steroid responsive, but 70% of episodes treated with oral prednisone relapsed, particularly at doses <10 mg daily or within 2 months of cessation. Immunotherapy, including maintenance prednisone (P=0.0004), intravenous immunoglobulin, rituximab and mycophenolate, all reduced median ARRs on-treatment. Treatment failure rates were lower in patients on maintenance steroids (5%) compared with non-steroidal maintenance immunotherapy (38%) (P=0.016). 58% of patients experienced residual disability (average follow-up 61 months, visual loss in 24%). Patients with ON were less likely to have sustained disability defined by a final EDSS of ≥2 (OR 0.15, P=0.032), while those who had any myelitis were more likely to have sustained residual deficits (OR 3.56, P=0.077).
Relapsing MOG antibody-associated demyelination is strongly associated with ON across all age groups and ADEM in children. Patients are highly responsive to steroids, but vulnerable to relapse on steroid reduction and cessation.
Symmetric Molecular Dynamics Cox, Sam; White, Andrew D.
Journal of chemical theory and computation,
07/2022, Volume:
18, Issue:
7
Journal Article
Peer reviewed
Open access
We derive a formulation of molecular dynamics that generates only symmetric configurations. We implement it for all 2D planar and 3D space groups. An atlas of 2D Lennard-Jones crystals under all ...planar groups is created with symmetric molecular dynamics.
Cysteine is commonly used to attach peptides onto gold surfaces. Here we show that the inclusion of an additional linker with a length of four residues (-PPPPC) and a rigid, hydrophobic nature is a ...better choice for forming peptide self-assembled monolayers (SAMs) with a well-ordered structure and high surface density. We compared the structure and function of the nonfouling peptide EKEKEKE-PPPPC-Am with EKEKEKE-C-Am. Circular dichroism, attenuated total internal reflection Fourier transform IR spectroscopy, and molecular dynamics results showed that EKEKEKE-PPPPC-Am forms a secondary structure while EKEKEKE-C-Am has a random structure. Surface plasmon resonance sensor results showed that protein adsorption on EKEKEKE-PPPPC-Am/gold is very low with small variation while protein adsorption on EKEKEKE-C-Am/gold is high with large variation. X-ray photoelectron spectroscopy results showed that both peptides have strong gold–thiol binding with the gold surface, indicating that their difference in protein adsorption is due to their assembled structures. Further experimental and simulation studies were performed to show that -PPPPC is a better linker than -PC, -PPC, and -PPPC. Finally, we extended EKEKEKE-PPPPC-Am with the cell-binding sequence RGD and demonstrated control over specific versus nonspecific cell adhesion without using poly(ethylene glycol). Adding a functional peptide to the nonfouling EK sequence avoids complex chemistries that are used for its connection to synthetic materials.