RESUMEN
BACKGROUND: Machine learning (ML) has a rich history in structural bioinformatics, and modern approaches, such as deep learning, are revolutionizing our knowledge of the subtle relationships between biomolecular sequence, structure, function, dynamics and evolution. As with any advance that rests upon statistical learning approaches, the recent progress in biomolecular sciences is enabled by the availability of vast volumes of sufficiently-variable data. To be useful, such data must be well-structured, machine-readable, intelligible and manipulable. These and related requirements pose challenges that become especially acute at the computational scales typical in ML. Furthermore, in structural bioinformatics such data generally relate to protein three-dimensional (3D) structures, which are inherently more complex than sequence-based data. A significant and recurring challenge concerns the creation of large, high-quality, openly-accessible datasets that can be used for specific training and benchmarking tasks in ML pipelines for predictive modeling projects, along with reproducible splits for training and testing. RESULTS: Here, we report 'Prop3D', a platform that allows for the creation, sharing and extensible reuse of libraries of protein domains, featurized with biophysical and evolutionary properties that can range from detailed, atomically-resolved physicochemical quantities (e.g., electrostatics) to coarser, residue-level features (e.g., phylogenetic conservation). As a community resource, we also supply a 'Prop3D-20sf' protein dataset, obtained by applying our approach to CATH . We have developed and deployed the Prop3D framework, both in the cloud and on local HPC resources, to systematically and reproducibly create comprehensive datasets via the Highly Scalable Data Service ( HSDS ). Our datasets are freely accessible via a public HSDS instance, or they can be used with accompanying Python wrappers for popular ML frameworks. CONCLUSION: Prop3D and its associated Prop3D-20sf dataset can be of broad utility in at least three ways. Firstly, the Prop3D workflow code can be customized and deployed on various cloud-based compute platforms, with scalability achieved largely by saving the results to distributed HDF5 files via HSDS . Secondly, the linked Prop3D-20sf dataset provides a hand-crafted, already-featurized dataset of protein domains for 20 highly-populated CATH families; importantly, provision of this pre-computed resource can aid the more efficient development (and reproducible deployment) of ML pipelines. Thirdly, Prop3D-20sf's construction explicitly takes into account (in creating datasets and data-splits) the enigma of 'data leakage', stemming from the evolutionary relationships between proteins.
Asunto(s)
Biología Computacional , Proteínas , Humanos , Filogenia , Biología Computacional/métodos , Flujo de Trabajo , Aprendizaje AutomáticoRESUMEN
This Special Issue of Biomolecules[...].
RESUMEN
Systematically discovering protein-ligand interactions across the entire human and pathogen genomes is critical in chemical genomics, protein function prediction, drug discovery, and many other areas. However, more than 90% of gene families remain "dark"-i.e., their small-molecule ligands are undiscovered due to experimental limitations or human/historical biases. Existing computational approaches typically fail when the dark protein differs from those with known ligands. To address this challenge, we have developed a deep learning framework, called PortalCG, which consists of four novel components: (i) a 3-dimensional ligand binding site enhanced sequence pre-training strategy to encode the evolutionary links between ligand-binding sites across gene families; (ii) an end-to-end pretraining-fine-tuning strategy to reduce the impact of inaccuracy of predicted structures on function predictions by recognizing the sequence-structure-function paradigm; (iii) a new out-of-cluster meta-learning algorithm that extracts and accumulates information learned from predicting ligands of distinct gene families (meta-data) and applies the meta-data to a dark gene family; and (iv) a stress model selection step, using different gene families in the test data from those in the training and development data sets to facilitate model deployment in a real-world scenario. In extensive and rigorous benchmark experiments, PortalCG considerably outperformed state-of-the-art techniques of machine learning and protein-ligand docking when applied to dark gene families, and demonstrated its generalization power for target identifications and compound screenings under out-of-distribution (OOD) scenarios. Furthermore, in an external validation for the multi-target compound screening, the performance of PortalCG surpassed the rational design from medicinal chemists. Our results also suggest that a differentiable sequence-structure-function deep learning framework, where protein structural information serves as an intermediate layer, could be superior to conventional methodology where predicted protein structures were used for the compound screening. We applied PortalCG to two case studies to exemplify its potential in drug discovery: designing selective dual-antagonists of dopamine receptors for the treatment of opioid use disorder (OUD), and illuminating the understudied human genome for target diseases that do not yet have effective and safe therapeutics. Our results suggested that PortalCG is a viable solution to the OOD problem in exploring understudied regions of protein functional space.
Asunto(s)
Algoritmos , Proteínas , Humanos , Ligandos , Proteínas/química , Sitios de Unión , Aprendizaje Automático , Unión ProteicaRESUMEN
Does reductionism, in the era of machine learning and now interpretable AI, facilitate or hinder scientific insight? The protein ribbon diagram, as a means of visual reductionism, is a case in point.
Asunto(s)
Aprendizaje Automático , SinapsisRESUMEN
Background: The COVID-19 pandemic is being battled via the largest vaccination campaign in history, with more than eight billion doses administered thus far. Therefore, discussions about potentially adverse reactions, and broader safety concerns, are critical. The U.S. Vaccination Adverse Event Reporting System (VAERS) has recorded vaccination side effects for over 30 years. About 580,000 events have been filed for COVID-19 thus far, primarily for the Johnson & Johnson (New Jersey, USA), Pfizer/BioNTech (Mainz, Germany), and Moderna (Cambridge, USA) vaccines. Methods: Using available databases, we evaluated these three vaccines in terms of the occurrence of four generally-noticed adverse reactionsnamely, cerebral venous sinus thrombosis, Guillain−Barré syndrome (a severe paralytic neuropathy), myocarditis, and pericarditis. Our statistical analysis also included a calculation of odds ratios (ORs) based on total vaccination numbers, accounting for incidence rates in the general population. Results: ORs for a number of adverse events and patient groups were (largely) increased, most notably for the occurrence of cerebral venous sinus thrombosis after vaccination with the Johnson & Johnson vaccine. The overall population OR of 10 increases to 12.5 when limited to women, and further yet (to 14.4) among women below age 50 yrs. In addition, elevated risks were found (i) for Guillain−Barré syndrome (OR of 11.6) and (ii) for myocarditis/pericarditis (ORs of 5.3/4.1, respectively) among young men (<25 yrs) vaccinated with the Pfizer/BioNTech vaccine. Conclusions: Any conclusions from such a retrospective, real-world data analysis must be drawn cautiously, and should be confirmed by prospective double-blinded clinical trials. In addition, we emphasize that the adverse events reported here are not specific side effects of COVID vaccines, and the significant, well-established benefits of COVID-19 vaccination outweigh the potential complications surveyed here.
RESUMEN
Advances in biomedicine are largely fueled by exploring uncharted territories of human biology. Machine learning can both enable and accelerate discovery, but faces a fundamental hurdle when applied to unseen data with distributions that differ from previously observed ones-a common dilemma in scientific inquiry. We have developed a new deep learning framework, called Portal Learning, to explore dark chemical and biological space. Three key, novel components of our approach include: (i) end-to-end, step-wise transfer learning, in recognition of biology's sequence-structure-function paradigm, (ii) out-of-cluster meta-learning, and (iii) stress model selection. Portal Learning provides a practical solution to the out-of-distribution (OOD) problem in statistical machine learning. Here, we have implemented Portal Learning to predict chemical-protein interactions on a genome-wide scale. Systematic studies demonstrate that Portal Learning can effectively assign ligands to unexplored gene families (unknown functions), versus existing state-of-the-art methods. Compared with AlphaFold2-based protein-ligand docking, Portal Learning significantly improved the performance by 79% in PR-AUC and 27% in ROC-AUC, respectively. The superior performance of Portal Learning allowed us to target previously "undruggable" proteins and design novel polypharmacological agents for disrupting interactions between SARS-CoV-2 and human proteins. Portal Learning is general-purpose and can be further applied to other areas of scientific inquiry.
RESUMEN
This Perspective examines a recent surge of information regarding the potential benefits of acid-suppression drugs in the context of COVID-19, with a particular eye on the great variability (and, thus, confusion) that has arisen across the reported findings, at least as regards the popular antacid famotidine. The degree of inconsistency and discordance reflects contradictory conclusions from independent, clinical-based studies that took roughly similar approaches, in terms of both experimental design (retrospective, observational, cohort-based, etc.) and statistical analysis workflows (propensity-score matching and stratification into sub-cohorts, etc.). The contradictions and potential confusion have ramifications for clinicians faced with choosing therapeutically optimal courses of intervention: e.g., do any potential benefits of famotidine suggest its use in a particular COVID-19 case? (If so, what administration route, dosage regimen, duration, etc. are likely optimal?) As succinctly put this March in Freedberg et al. (2021), " several retrospective studies show relationships between famotidine and outcomes in COVID-19 and several do not." Beyond the pressing issue of possible therapeutic indications, the conflicting data and conclusions related to famotidine must be resolved before its inclusion/integration in ontological and knowledge graph (KG)-based frameworks, which in turn are useful for drug discovery and repurposing. As a broader methodological issue, note that reconciling inconsistencies would bolster the validity of meta-analyses which draw upon the relevant data-sources. And, perhaps most broadly, developing a system for treating inconsistencies would stand to improve the qualities of both 1) real world evidence-based studies (retrospective), on the one hand, and 2) placebo-controlled, randomized multi-center clinical trials (prospective), on the other hand. In other words, a systematic approach to reconciling the two types of studies would inherently improve the quality and utility of each type of study individually.
RESUMEN
Immune dysregulation is characteristic of the more severe stages of SARS-CoV-2 infection. Understanding the mechanisms by which the immune system contributes to COVID-19 severity may open new avenues to treatment. Here, we report that elevated IL-13 was associated with the need for mechanical ventilation in 2 independent patient cohorts. In addition, patients who acquired COVID-19 while prescribed Dupilumab, a mAb that blocks IL-13 and IL-4 signaling, had less severe disease. In SARS-CoV-2-infected mice, IL-13 neutralization reduced death and disease severity without affecting viral load, demonstrating an immunopathogenic role for this cytokine. Following anti-IL-13 treatment in infected mice, hyaluronan synthase 1 (Has1) was the most downregulated gene, and accumulation of the hyaluronan (HA) polysaccharide was decreased in the lung. In patients with COVID-19, HA was increased in the lungs and plasma. Blockade of the HA receptor, CD44, reduced mortality in infected mice, supporting the importance of HA as a pathogenic mediator. Finally, HA was directly induced in the lungs of mice by administration of IL-13, indicating a new role for IL-13 in lung disease. Understanding the role of IL-13 and HA has important implications for therapy of COVID-19 and, potentially, other pulmonary diseases. IL-13 levels were elevated in patients with severe COVID-19. In a mouse model of the disease, IL-13 neutralization reduced the disease and decreased lung HA deposition. Administration of IL-13-induced HA in the lung. Blockade of the HA receptor CD44 prevented mortality, highlighting a potentially novel mechanism for IL-13-mediated HA synthesis in pulmonary pathology.
Asunto(s)
COVID-19/inmunología , Interleucina-13/inmunología , SARS-CoV-2/inmunología , Animales , COVID-19/sangre , COVID-19/patología , COVID-19/terapia , Modelos Animales de Enfermedad , Progresión de la Enfermedad , Femenino , Humanos , Interleucina-13/sangre , Pulmón/inmunología , Pulmón/patología , Masculino , Ratones , Ratones Endogámicos C57BL , Índice de Severidad de la EnfermedadRESUMEN
Immune dysregulation is characteristic of the more severe stages of SARS-CoV-2 infection. Understanding the mechanisms by which the immune system contributes to COVID-19 severity may open new avenues to treatment. Here we report that elevated interleukin-13 (IL-13) was associated with the need for mechanical ventilation in two independent patient cohorts. In addition, patients who acquired COVID-19 while prescribed Dupilumab had less severe disease. In SARS-CoV-2 infected mice, IL-13 neutralization reduced death and disease severity without affecting viral load, demonstrating an immunopathogenic role for this cytokine. Following anti-IL-13 treatment in infected mice, in the lung, hyaluronan synthase 1 (Has1) was the most downregulated gene and hyaluronan accumulation was decreased. Blockade of the hyaluronan receptor, CD44, reduced mortality in infected mice, supporting the importance of hyaluronan as a pathogenic mediator, and indicating a new role for IL-13 in lung disease. Understanding the role of IL-13 and hyaluronan has important implications for therapy of COVID-19 and potentially other pulmonary diseases.
Asunto(s)
Biología Computacional , Investigación , Adolescente , Biología Computacional/educación , Humanos , Laboratorios , Tutoría , MentoresRESUMEN
BACKGROUND: Given that an individual's age and gender are strongly predictive of coronavirus disease 2019 (COVID-19) outcomes, do such factors imply anything about preferable therapeutic options? METHODS: An analysis of electronic health records for a large (68,466-case), international COVID-19 cohort, in 5-year age strata, revealed age-dependent sex differences. In particular, we surveyed the effects of systemic hormone administration in women. The primary outcome for estradiol therapy was death. Odds ratios (ORs) and Kaplan-Meier survival curves were analyzed for 37,086 COVID-19 women in two age groups: pre- (15-49 years) and peri-/post-menopausal (> 50 years). RESULTS: The incidence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is higher in women than men (by about + 15%) and, in contrast, the fatality rate is higher in men (about + 50%). Interestingly, the relationships between these quantities are linked to age: pre-adolescent girls and boys had the same risk of infection and fatality rate, while adult premenopausal women had a significantly higher risk of infection than men in the same 5-year age stratum (about 16,000 vs. 12,000 cases). This ratio changed again in peri- and postmenopausal women, with infection susceptibility converging with men. While fatality rates increased continuously with age for both sexes, at 50 years, there was a steeper increase for men. Thus far, these types of intricacies have been largely neglected. Because the hormone 17ß-estradiol influences expression of the human angiotensin-converting enzyme 2 (ACE2) protein, which plays a role in SARS-CoV-2 cellular entry, propensity score matching was performed for the women's sub-cohort, comparing users vs. non-users of estradiol. This retrospective study of hormone therapy in female COVID-19 patients shows that the fatality risk for women > 50 years receiving estradiol therapy (user group) is reduced by more than 50%; the OR was 0.33, 95% CI [0.18, 0.62] and the hazard ratio (HR) was 0.29, 95% CI [0.11,0.76]. For younger, pre-menopausal women (15-49 years), the risk of COVID-19 fatality is the same irrespective of estradiol treatment, probably because of higher endogenous estradiol levels. CONCLUSIONS: As of this writing, still no effective drug treatment is available for COVID-19; since estradiol shows such a strong improvement regarding fatality in COVID-19, we suggest prospective studies on the potentially more broadly protective roles of this naturally occurring hormone.
Asunto(s)
COVID-19/epidemiología , Estradiol/uso terapéutico , Peptidil-Dipeptidasa A/uso terapéutico , Neumonía Viral/epidemiología , Adolescente , Adulto , COVID-19/prevención & control , Femenino , Humanos , Masculino , Persona de Mediana Edad , Neumonía Viral/tratamiento farmacológico , Estudios Retrospectivos , SARS-CoV-2 , Caracteres Sexuales , Adulto JovenRESUMEN
The rapidly developing pandemic, known as coronavirus disease 2019 (COVID-19) and caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has recently spread across 213 countries and territories. This pandemic is a dire public health threat-particularly for those suffering from hypertension, cardiovascular diseases, pulmonary diseases, or diabetes; without approved treatments, it is likely to persist or recur. To facilitate the rapid discovery of inhibitors with clinical potential, we have applied ligand- and structure-based computational approaches to develop a virtual screening methodology that allows us to predict potential inhibitors. In this work, virtual screening was performed against two natural products databases, Super Natural II and Traditional Chinese Medicine. Additionally, we have used an integrated drug repurposing approach to computationally identify potential inhibitors of the main protease of SARS-CoV-2 in databases of drugs (both approved and withdrawn). Roughly 360,000 compounds were screened using various molecular fingerprints and molecular docking methods; of these, 80 docked compounds were evaluated in detail, and the 12 best hits from four datasets were further inspected via molecular dynamics simulations. Finally, toxicity and cytochrome inhibition profiles were computationally analyzed for the selected candidate compounds.
RESUMEN
We suspect that there is a level of granularity of protein structure intermediate between the classical levels of "architecture" and "topology," as reflected in such phenomena as extensive three-dimensional structural similarity above the level of (super)folds. Here, we examine this notion of architectural identity despite topological variability, starting with a concept that we call the "Urfold." We believe that this model could offer a new conceptual approach for protein structural analysis and classification: indeed, the Urfold concept may help reconcile various phenomena that have been frequently recognized or debated for years, such as the precise meaning of "significant" structural overlap and the degree of continuity of fold space. More broadly, the role of structural similarity in sequenceâstructureâfunction evolution has been studied via many models over the years; by addressing a conceptual gap that we believe exists between the architecture and topology levels of structural classification schemes, the Urfold eventually may help synthesize these models into a generalized, consistent framework. Here, we begin by qualitatively introducing the concept.
Asunto(s)
Proteínas/química , Algoritmos , Modelos Moleculares , Conformación Proteica , Pliegue de ProteínaRESUMEN
Short peptides are uniquely versatile building blocks for self-assembly. Supramolecular peptide assemblies can be used to construct functional hydrogel biomaterials-an attractive approach for neural tissue engineering. Here, we report a new class of short, five-residue peptides that form hydrogels with nanofiber structures. Using rheology and spectroscopy, we describe how sequence variations, pH, and peptide concentration alter the mechanical properties of our pentapeptide hydrogels. We find that this class of seven unmodified peptides forms robust hydrogels from 0.2-20 kPa at low weight percent (less than 3 wt %) in cell culture media and undergoes shear-thinning and rapid self-healing. The peptides self-assemble into long fibrils with sequence-dependent fibrillar morphologies. These fibrils exhibit a unique twisted ribbon shape, as visualized by transmission electron microscopy (TEM) and Cryo-EM imaging, with diameters in the low tens of nanometers and periodicities similar to amyloid fibrils. Experimental gelation behavior corroborates our molecular dynamics simulations, which demonstrate peptide assembly behavior, an increase in ß-sheet content, and patterns of variation in solvent accessibility. Our rapidly assembling pentapeptides for injectable delivery (RAPID) hydrogels are syringe-injectable and support cytocompatible encapsulation of oligodendrocyte progenitor cells (OPCs), as well as their proliferation and three-dimensional process extension. Furthermore, RAPID gels protect OPCs from mechanical membrane disruption and acute loss of viability when ejected from a syringe needle, highlighting the protective capability of the hydrogel as potential cell carriers for transplantation therapies. The tunable mechanical and structural properties of these supramolecular assemblies are shown to be permissive to cell expansion and remodeling, making this hydrogel system suitable as an injectable material for cell delivery and tissue engineering applications.
Asunto(s)
Materiales Biocompatibles/química , Materiales Biocompatibles/farmacología , Hidrogeles/química , Nanofibras/química , Oligopéptidos/química , Ingeniería de Tejidos , Secuencia de Aminoácidos , Encéfalo/citología , Encéfalo/efectos de los fármacos , Concentración de Iones de Hidrógeno , Fenómenos Mecánicos , Simulación de Dinámica Molecular , Estructura Secundaria de Proteína , ReologíaRESUMEN
The small ß-barrel (SBB) is an ancient protein structural domain characterized by extremes: it features a broad range of structural varieties, a deeply intricate evolutionary history, and it is associated with a bewildering array of cellular pathways. Here, we present a thorough, survey-based analysis of the structural properties of SBBs. We first consider the defining properties of the SBB, including various systems of nomenclature used to describe it, and we introduce the unifying concept of an "urfold." To begin elucidating how vast functional diversity can be achieved by a relatively simple domain, we explore the anatomy of the SBB and its representative structural variants. Many SBB proteins assemble into cyclic oligomers as the biologically functional units; these oligomers often bind RNA, and typically exhibit great quaternary structural plasticity (homomeric and heteromeric rings, variable subunit stoichiometries, etc.). We conclude with three themes that emerge from the rich structure â function versatility of the SBB.
Asunto(s)
Proteínas/química , Animales , Sitios de Unión , Humanos , Modelos Moleculares , Unión Proteica , Estructura Secundaria de ProteínaRESUMEN
Data science has emerged from the proliferation of digital data, coupled with advances in algorithms, software and hardware (e.g., GPU computing). Innovations in structural biology have been driven by similar factors, spurring us to ask: can these two fields impact one another in deep and hitherto unforeseen ways? We posit that the answer is yes. New biological knowledge lies in the relationships between sequence, structure, function and disease, all of which play out on the stage of evolution, and data science enables us to elucidate these relationships at scale. Here, we consider the above question from the five key pillars of data science: acquisition, engineering, analytics, visualization and policy, with an emphasis on machine learning as the premier analytics approach.