OpenZika: An IBM World Community Grid Project to Accelerate Zika Virus Drug Discovery

The Zika virus outbreak in the Americas has caused global concern. To help accelerate this fight against Zika, we launched the OpenZika project. OpenZika is an IBM World Community Grid Project that uses distributed computing on millions of computers and Android devices to run docking experiments, in order to dock tens of millions of drug-like compounds against crystal structures and homology models of Zika proteins (and other related flavivirus targets). This will enable the identification of new candidates that can then be tested in vitro, to advance the discovery and development of new antiviral drugs against the Zika virus. The docking data is being made openly accessible so that all members of the global research community can use it to further advance drug discovery studies against Zika and other related flaviviruses.The Zika virus (ZIKV) has emerged as a major public health threat to the Americas as of 2015 [1]. We have previously suggested that it represents an opportunity for scientific collaboration and open scientific exchange [2]. The health of future generations may very well depend on the decisions we make, our willingness to share our findings quickly, and open collaboration to rapidly find a cure for this disease. Since February 1, 2016, when the World Health Organization deemed the cluster of microcephaly cases, Guillain-Barré, and other neurological disorders associated with ZIKV in Latin America and the Caribbean as constituting a Public Health Emergency of International Concern [3] (PHEIC), we have seen a rapid increase in publications (S1 References and main references). We [2] and others [4,5] described steps that could be taken to initiate a drug discovery program on ZIKV. For example, computational approaches, such as virtual screening of chemical libraries or focused screening to repurpose FDA and/or EU-approved drugs, can be used to help accelerate the discovery of an anti-ZIKV drug. An antiviral drug discovery program can be initiated using structure-based design, based on homology models of the key ZIKV proteins. With the lack of structural information regarding the proteins of ZIKV, we built homology models for all the ZIKV proteins, based on close homologs such as dengue virus, using freely available software [6] (S1 Table). These were made available online on March 3, 2016. We also predicted the site of glycosylation of glycoprotein E as Asn154, which was recently experimentally verified [7].Since the end of March 2016, we have now seen two cryo-EM structures and 16 crystal structures of five target classes (S1 Table). These structures, alongside the homology models, represent potential starting points for docking-based virtual screening campaigns to help find molecules that are predicted to have high affinity with ZIKV proteins. These predictions can then be tested against the virus in cell-based assays and/or using individual protein-based assays. There are millions of molecules available that can be assayed, but which ones are likely to work, and how should we prioritize them?In March, we initiated a new open collaborative project called OpenZika (Fig 1), with IBM’s World Community Grid (WCG, worldcommunitygrid.org), which has been used previously for distributed computing projects (S2 Table). On May 18, 2016, the OpenZika project began the virtual screening of ~6 million compounds that are in the ZINC database (Fig 1), as well as the FDA-approved drugs and the NIH clinical collection, using AutoDock Vina and the homology models and crystal structures (S1 Table, S1 Text, S1 References), to discover novel candidate compounds that can potentially be developed into new drugs for treating ZIKV. These will be followed by additional virtual screens with a new ZINC library of ~38 million compounds, and the PubChem database (at most ~90 million compounds), after their structures are prepared for docking.Open in a separate windowFig 1Workflow for the OpenZika project.A. Docking input files of the targets and ligands are prepared, and positive control docking studies are performed. The crystallographic binding mode of a known inhibitor is shown as sticks with dark purple carbon atoms, while the docked binding mode against the NS5 target from HCV has cyan carbons. Our pdbqt files of the libraries of compounds we screen are also openly accessible (http://zinc.docking.org/pdbqt/). B. We have already prepared the docking input files for ~6 million compounds from ZINC (i.e., the libraries that ALP previously used in the GO Fight Against Malaria project on World Community Grid), which are currently being used in the initial set of virtual screens on OpenZika. C. IBM’s World Community Grid is an internet-distributed network of millions of computers (Mac, Windows, and Linux) and Android-based tablets or smartphones in over 80 countries. Over 715,000 volunteers donate their dormant computer time (that would otherwise be wasted) towards different projects that are both (a) run by an academic or nonprofit research institute, and (b) are devoted to benefiting humanity. D. OpenZika is harnessing World Community Grid to dock millions of commercially available compounds against multiple ZIKV homology models and crystal structures (and targets from related viruses) using AutoDock Vina (AD Vina). This ultimately produces candidates (virtual hits that produced the best docking scores and displayed the best interactions with the target during visual inspection) against individual proteins, which can then be prioritized for in vitro testing by collaborators. After it is inspected, all computational data against ZIKV targets will be made open to the public on our website (http://openzika.ufg.br/experiments/#tab-id-7), and OpenZika results are also available upon request. The computational and experimental data produced will be published as quickly as possible.Initially, compounds are being screened against the ZIKV homologs of drug targets that have been well-validated in research against dengue and hepatitis C viruses, such as NS5 and Glycoprotein E (S1 Table, S1 Text, S1 References). These may allow us to identify broad-spectrum antivirals against multiple flaviviruses, such as dengue virus, West Nile virus, and yellow fever virus. In addition, docking against the crystal structure of a related protein from a different pathogen can sometimes discover novel hits against the pathogen of interest [8].As well as applying docking-based filters, the compounds virtually screened on OpenZika will also be filtered using machine learning models (S1 Text, S1 References). These should be useful selection criteria for subsequent tests by our collaborators in whole-cell ZIKV assays, to verify their antiviral activity for blocking ZIKV infection or replication. Since all OpenZika docking data will be in the public domain soon after they are completed and verified, we and other labs can then advance the development of some of these new virtual candidates into experimentally validated hits, leads, and drugs through collaborations with wet labs.This exemplifies open science, which should help scientists around the world as they address the long and arduous process of discovering and developing new drugs. Screening millions of compounds against many different protein models in this way would take far more resources and time than any academic researcher could generally obtain or spend. As of August 16, 2016, we have submitted 894 million docking jobs. Over 6,934 CPU years have been donated to us, enabling over 439 million different docking jobs. We recently selected an initial batch of candidates for NS3 helicase (data openly available at http://openzika.ufg.br/experiments/#tab-id-7), for in vitro testing. Without the unique community of volunteers and tremendous resources provided by World Community Grid, this project would have been very difficult to initiate in a reasonable time frame at this scale.The OpenZika project will ultimately generate several billion docking results, which could make it the largest computational drug discovery project ever performed in academia. The potential challenges we foresee will be finding laboratories with sufficient funding to pursue compounds, synthesize analogs, and develop target-based assays to validate our predictions and generate SAR (Structure-Activity Relationship) data to guide the process of developing the new hits into leads and then drugs. Due to the difficult nature of drug discovery and the eventual evolution of drug resistance, funding of ZIKV research once initiated will likely need to be sustained for several years, if not longer (e.g., HIV research has been funded for decades). As with other WCG projects, once scientists identify experimentally validated leads, finding a company to license them and pursue them in clinical trials and beyond will need incentives such as the FDA Tropical Disease Priority voucher, [9] which has a financial value on the open market [10].By working together and opening our research to the scientific community, many other labs will also be able to take promising molecular candidates forward to accelerate progress towards defeating the ZIKV outbreak. We invite any interested researcher to join us (send us your models or volunteer to assay the candidates we identify through this effort against any of the flaviviruses), and we hope new volunteers in the general public will donate their dormant, spare computing cycles to this cause. We will ultimately report the full computational and experimental results of this collaboration.

Advantages and Disadvantages of OpenZika

Advantages

• Open Science could accelerate the discovery of new antivirals using docking and virtual screening
• Docking narrows down compounds to test, which saves time and money
• Free to use distributed computing on World Community Grid, and the workflow is simpler than using conventional supercomputers

Disadvantages

• Concern around intellectual property ownership and whether companies will develop drugs coming from effort
• Need for experimental assays will always be a factor
• Testing in vitro and in vivo is not free, nor are the samples of the compounds

     

Paternal B Vitamin Intake Is a Determinant of Growth,Hepatic Lipid Metabolism and Intestinal Tumor Volume in Female Apc1638N Mouse Offspring

Background

The importance of maternal nutrition to offspring health and risk of disease is well established. Emerging evidence suggests paternal diet may affect offspring health as well.

Objective

In the current study we sought to determine whether modulating pre-conception paternal B vitamin intake alters intestinal tumor formation in offspring. Additionally, we sought to identify potential mechanisms for the observed weight differential among offspring by profiling hepatic gene expression and lipid content.

Methods

Male Apc^1638N mice (prone to intestinal tumor formation) were fed diets containing replete (control, CTRL), mildly deficient (DEF), or supplemental (SUPP) quantities of vitamins B₂, B₆, B₁₂, and folate for 8 weeks before mating with control-fed wild type females. Wild type offspring were euthanized at weaning and hepatic gene expression profiled. Apc^1638N offspring were fed a replete diet and euthanized at 28 weeks of age to assess tumor burden.

Results

No differences in intestinal tumor incidence or burden were found between male Apc^1638N offspring of different paternal diet groups. Although in female Apc^1638N offspring there were no differences in tumor incidence or multiplicity, a stepwise increase in tumor volume with increasing paternal B vitamin intake was observed. Interestingly, female offspring of SUPP and DEF fathers had a significantly lower body weight than those of CTRL fed fathers. Moreover, hepatic trigylcerides and cholesterol were elevated 3-fold in adult female offspring of SUPP fathers. Weanling offspring of the same fathers displayed altered expression of several key lipid-metabolism genes. Hundreds of differentially methylated regions were identified in the paternal sperm in response to DEF and SUPP diets. Aside from a few genes including Igf2, there was a striking lack of overlap between these genes differentially methylated in sperm and differentially expressed in offspring.

Conclusions

In this animal model, modulation of paternal B vitamin intake prior to mating alters offspring weight gain, lipid metabolism and tumor growth in a sex-specific fashion. These results highlight the need to better define how paternal nutrition affects the health of offspring.     

Identifying Likely Transmission Pathways within a 10-Year Community Outbreak of Tuberculosis by High-Depth Whole Genome Sequencing

Background

Improved tuberculosis control and the need to contain the spread of drug-resistant strains provide a strong rationale for exploring tuberculosis transmission dynamics at the population level. Whole-genome sequencing provides optimal strain resolution, facilitating detailed mapping of potential transmission pathways.

Methods

We sequenced 22 isolates from a Mycobacterium tuberculosis cluster in New South Wales, Australia, identified during routine 24-locus mycobacterial interspersed repetitive unit typing. Following high-depth paired-end sequencing using the Illumina HiSeq 2000 platform, two independent pipelines were employed for analysis, both employing read mapping onto reference genomes as well as de novo assembly, to control biases in variant detection. In addition to single-nucleotide polymorphisms, the analyses also sought to identify insertions, deletions and structural variants.

Results

Isolates were highly similar, with a distance of 13 variants between the most distant members of the cluster. The most sensitive analysis classified the 22 isolates into 18 groups. Four of the isolates did not appear to share a recent common ancestor with the largest clade; another four isolates had an uncertain ancestral relationship with the largest clade.

Conclusion

Whole genome sequencing, with analysis of single-nucleotide polymorphisms, insertions, deletions, structural variants and subpopulations, enabled the highest possible level of discrimination between cluster members, clarifying likely transmission pathways and exposing the complexity of strain origin. The analysis provides a basis for targeted public health intervention and enhanced classification of future isolates linked to the cluster.     

Influence of Manatees' Diving on Their Risk of Collision with Watercraft

Watercraft pose a threat to endangered Florida manatees (Trichechus manatus latirostris). Mortality from watercraft collisions has adversely impacted the manatee population’s growth rate, therefore reducing this threat is an important management goal. To assess factors that contribute to the risk of watercraft strikes to manatees, we studied the diving behavior of nine manatees carrying GPS tags and time–depth recorders in Tampa Bay, Florida, during winters 2002–2006. We applied a Bayesian formulation of generalized linear mixed models to depth data to model the probability (P_t) that manatees would be no deeper than 1.25 m from the water’s surface as a function of behavioral and habitat covariates. Manatees above this threshold were considered to be within striking depth of a watercraft. Seventy-eight percent of depth records (individual range 62–86%) were within striking depth (mean = 1.09 m, max = 16.20 m), illustrating how vulnerable manatees are to strikes. In some circumstances manatees made consecutive dives to the bottom while traveling, even in areas >14 m, possibly to conserve energy. This is the first documentation of potential cost-efficient diving behavior in manatees. Manatees were at higher risk of being within striking depth in shallow water (<0.91 m), over seagrass, at night, and while stationary or moving slowly; they were less likely to be within striking depth when ≤50 m from a charted waterway. In shallow water the probability of a manatee being within striking depth was 0.96 (CI = 0.93–0.98) and decreased as water depth increased. The probability was greater over seagrass (P_t = 0.96, CI = 0.93–0.98) than over other substrates (P_t = 0.73, CI = 0.58–0.84). Quantitative approaches to assessing risk can improve the effectiveness of manatee conservation measures by helping identify areas for protection.     

The Ancestral Carnivore Karyotype As Substantiated by Comparative Chromosome Painting of Three Pinnipeds,the Walrus,the Steller Sea Lion and the Baikal Seal (Pinnipedia,Carnivora)

Karyotype evolution in Carnivora is thoroughly studied by classical and molecular cytogenetics and supplemented by reconstructions of Ancestral Carnivora Karyotype (ACK). However chromosome painting information from two pinniped families (Odobenidae and Otariidae) is noticeably missing. We report on the construction of the comparative chromosome map for species from each of the three pinniped families: the walrus (Odobenus rosmarus, Odobenidae–monotypic family), near threatened Steller sea lion (Eumetopias jubatus, Otariidae) and the endemic Baikal seal (Pusa sibirica, Phocidae) using combination of human, domestic dog and stone marten whole-chromosome painting probes. The earliest karyological studies of Pinnipedia showed that pinnipeds were characterized by a pronounced karyological conservatism that is confirmed here with species from Phocidae, Otariidae and Odobenidae sharing same low number of conserved human autosomal segments (32). Chromosome painting in Pinnipedia and comparison with non-pinniped carnivore karyotypes provide strong support for refined structure of ACK with 2n = 38. Constructed comparative chromosome maps show that pinniped karyotype evolution was characterized by few tandem fusions, seemingly absent inversions and slow rate of genome rearrangements (less then one rearrangement per 10 million years). Integrative comparative analyses with published chromosome painting of Phoca vitulina revealed common cytogenetic signature for Phoca/Pusa branch and supports Phocidae and Otaroidea (Otariidae/Odobenidae) as sister groups. We revealed rearrangements specific for walrus karyotype and found the chromosomal signature linking together families Otariidae and Odobenidae. The Steller sea lion karyotype is the most conserved among three studied species and differs from the ACK by single fusion. The study underlined the strikingly slow karyotype evolution of the Pinnipedia in general and the Otariidae in particular.     

Ligation of Glycophorin A Generates Reactive Oxygen Species Leading to Decreased Red Blood Cell Function

Acute, inflammatory conditions associated with dysregulated complement activation are characterized by significant increases in blood concentration of reactive oxygen species (ROS) and ATP. The mechanisms by which these molecules arise are not fully understood. In this study, using luminometric- and fluorescence-based methods, we show that ligation of glycophorin A (GPA) on human red blood cells (RBCs) results in a 2.1-fold, NADPH-oxidase-dependent increase in intracellular ROS that, in turn, trigger multiple downstream cascades leading to caspase-3 activation, ATP release, and increased band 3 phosphorylation. Functionally, using 2D microchannels to assess membrane deformability, GPS-ligated RBCs travel 33% slower than control RBCs, and lipid mobility was hindered by 10% using fluorescence recovery after photobleaching (FRAP). These outcomes were preventable by pretreating RBCs with cell-permeable ROS scavenger glutathione monoethyl ester (GSH-ME). Our results obtained in vitro using anti-GPA antibodies were validated using complement-altered RBCs isolated from control and septic patients. Our results suggest that during inflammatory conditions, circulating RBCs significantly contribute to capillary flow dysfunctions, and constitute an important but overlooked source of intravascular ROS and ATP, both critical mediators responsible for endothelial cell activation, microcirculation impairment, platelet activation, as well as long-term dysregulated adaptive and innate immune responses.     

Changes in Colonic Bile Acid Composition following Fecal Microbiota Transplantation Are Sufficient to Control Clostridium difficile Germination and Growth

Fecal microbiota transplantation (FMT) is a highly effective therapy for recurrent Clostridium difficile infection (R-CDI), but its mechanisms remain poorly understood. Emerging evidence suggests that gut bile acids have significant influence on the physiology of C. difficile, and therefore on patient susceptibility to recurrent infection. We analyzed spore germination of 10 clinical C. difficile isolates exposed to combinations of bile acids present in patient feces before and after FMT. Bile acids at concentrations found in patients’ feces prior to FMT induced germination of C. difficile, although with variable potency across different strains. However, bile acids at concentrations found in patients after FMT did not induce germination and inhibited vegetative growth of all C. difficile strains. Sequencing of the newly identified germinant receptor in C. difficile, CspC, revealed a possible correspondence of variation in germination responses across isolates with mutations in this receptor. This may be related to interstrain variability in spore germination and vegetative growth in response to bile acids seen in this and other studies. These results support the idea that intra-colonic bile acids play a key mechanistic role in the success of FMT, and suggests that novel therapeutic alternatives for treatment of R-CDI may be developed by targeted manipulation of bile acid composition in the colon.     

Learning Biomarker Models for Progression Estimation of Alzheimer’s Disease

Being able to estimate a patient’s progress in the course of Alzheimer’s disease and predicting future progression based on a number of observed biomarker values is of great interest for patients, clinicians and researchers alike. In this work, an approach for disease progress estimation is presented. Based on a set of subjects that convert to a more severe disease stage during the study, models that describe typical trajectories of biomarker values in the course of disease are learned using quantile regression. A novel probabilistic method is then derived to estimate the current disease progress as well as the rate of progression of an individual by fitting acquired biomarkers to the models. A particular strength of the method is its ability to naturally handle missing data. This means, it is applicable even if individual biomarker measurements are missing for a subject without requiring a retraining of the model. The functionality of the presented method is demonstrated using synthetic and—employing cognitive scores and image-based biomarkers—real data from the ADNI study. Further, three possible applications for progress estimation are demonstrated to underline the versatility of the approach: classification, construction of a spatio-temporal disease progression atlas and prediction of future disease progression.