key: cord-0697972-pajqblus authors: Hoshino, Kunikazu; Maeshiro, Tatsuji; Nishida, Nao; Sugiyama, Masaya; Fujita, Jiro; Gojobori, Takashi; Mizokami, Masashi title: Transmission dynamics of SARS-CoV-2 on the Diamond Princess uncovered using viral genome sequence analysis date: 2021-02-13 journal: Gene DOI: 10.1016/j.gene.2021.145496 sha: 22d20400548ad9bde17efefedda651f603c0d9b0 doc_id: 697972 cord_uid: pajqblus An outbreak of coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) occurred aboard the Diamond Princess cruise ship between her January 20 departure and late February 2020. Here, we used phylodynamic analyses to investigate the transmission dynamics of SARS-CoV-2 during the outbreak. Using a Bayesian coalescent-based method, the estimated mean nucleotide substitution rate of 240 SARS-CoV-2 whole-genome sequences was approximately 7.13 × 10−4 substitutions per site per year. Population dynamics and the effective reproductive number (Re) of SARS-CoV-2 infections were estimated using a Bayesian framework. The estimated origin of the outbreak was January 21, 2020. The infection spread substantially before quarantine on February 5. The Re peaked at 6.06 on February 4 and gradually declined to 1.51, suggesting that transmission continued slowly even after quarantine. These findings highlight the high transmissibility of SARS-CoV-2 and the need for effective measures to control outbreaks in confined settings. Coronavirus disease 2019 (COVID-19) is a respiratory infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (Coronaviridae Study Group of the International Committee on Taxonomy of Viruses, 2020; . The initial case was reported in China in December 2019 , and the outbreak spread rapidly around the world resulting in an ongoing pandemic (WHO, 2020b) . As of September 27, 2020, the total number of confirmed cases had reached 32.7 million, and 991,000 people had died of COVID-19 (WHO, 2020a) . Coronaviruses are a family of enveloped positive-strand RNA viruses that possess a single-stranded, positive-sense RNA genome comprising 26-32 kilobases (kb). Among them, SARS-CoV-2 belongs to the genus Betacoronavirus and has a genome that is approximately 30 kb in length (Lu et al., 2020) . Some betacoronaviruses have caused severe disease outbreaks such as severe acute respiratory syndrome (SARS) and the Middle East respiratory syndrome (MERS). The COVID-19 pandemic is the third major outbreak caused by betacoronaviruses in the past two decades. In contrast to SARS and MERS, a substantial proportion of SARS-CoV-2 infections causing COVID-19 have been asymptomatic (He et al., 2020; Nishiura et al., 2020) . CoV-2 can transmit the virus (Bai et al., 2020; Ganyani et al., 2020; Rothe et al., 2020) , epidemiological data can be unreliable, causing difficulties in understanding transmission dynamics and controlling the spread of infection. An outbreak of COVID-19 occurred on board the Diamond Princess cruise ship in February 2020. The Diamond Princess sailed from Yokohama on January 20, 2020, arrived at Hong Kong on January 25, and returned to Yokohama on February 3 with 2,666 passengers and 1,045 crew members. The presumed COVID-19 index case was reportedly a passenger from Hong Kong who boarded the ship in Yokohama on January 20 and developed a mild dry cough on January 23 . After the passenger returned to Hong Kong and disembarked on January 25, he became feverish on January 29 and was diagnosed with COVID-19 on February 1. After learning that a former passenger had been diagnosed with COVID-19, the Japanese Ministry of Health, Labour, and Welfare quarantined the ship in Yokohama on February 5. Movement of the passengers was restricted, and self-isolation in cabins was implemented. Symptomatic and high-risk individuals were initially tested for SARS-CoV-2 using reverse-transcription polymerase chain reaction (RT-PCR), and then, asymptomatic passengers were tested after February 11 (National institute of infectious diseases, 2020; Nishiura, 2020; Yamahata and Shibata, 2020) . Individuals testing positive were Hoshino et al. 5 transported from the ship to the quarantine location or hospitals (Yamahata and Shibata, 2020) . A total of 712 individuals had tested positive among > 3,600 tests as of August 2020 (Ministry of Health). Understanding the patterns of transmission and the dynamics of an infectious agent, among a susceptible population, is essential for establishing effective strategies to control and prevent the spread of infection, particularly when the pathogenic agent is novel (Gardy and Loman, 2018; Ladner et al., 2019) . The COVID-19 outbreak on the Diamond Princess cruise ship might provide valuable samples and information about the transmission dynamics of SARS-CoV-2, particularly in a confined setting. In this case, the presumed index case is known, the number of susceptible individuals was constant, timely preventive measures were deployed, and the passengers were extensively tested using RT-PCR . Although the number of confirmed cases increased soon after returning to port suggesting that the infection had spread among passengers and crew during the voyage, the temporal dynamics of transmission remained elusive. Here, we analyzed the transmission dynamics of SARS-CoV-2 aboard the Diamond Princess using its viral genome sequence in a Bayesian framework. We estimated the effective population size, as well as basic (R 0 ) and effective (R e ) Hoshino et al. 6 reproductive numbers of SARS-CoV-2 during the outbreak and evaluated changes in transmissibility over time. All publicly available SARS-CoV-2 genome sequences with clinical information as of August 7, 2020 (n = 78,448), were retrieved from the Global Initiative on Sharing All Influenza Data (GISAID) database (Elbe and Buckland-Merrett, 2017) . We included sequences with specific collection date known to the day, sampling location, and host information indicating that they had been collected from human hosts. We retained only complete or nearly complete sequences between 29,600 and 31,000 nt that contained <1% ambiguous nucleotides. Sequences were aligned and trimmed according to the reference sequence (NC_045512.2, 29,903 nt) using MAFFT (v.7.453; https://mafft.cbrc.jp/alignment/software/) with the "-addfragments" option. Finally, we excluded sequences with >20 nt gaps, >20 mismatched nucleotides, or concentrated ambiguity calls (>10 in any 50-nt window) as described (Korber et al., 2020) and manually inspected the sequences using AliView (v.1.26; https://ormbunkar.se/aliview/). All retained sequences were randomly resampled up to Hoshino et al. 7 two per country, year, and month using nextstrain CLI (v.1.16.5) (Hadfield et al., 2018) to maintain the geographic and temporal diversity of the virus. All available isolates from Wuhan sampled in December, 2019 were included in the dataset. Figure 1 shows the characteristics of single nucleotide variants in the reference dataset. In total, 74 SARS-CoV-2 sequences from infected persons on the Diamond Princess were identified in the GISAID database as of August 7, 2020. Sequences with multiple ambiguous nucleotides (>1%) and long gaps in the coding region were excluded. Finally, 67 sequences were retained and subsequently analyzed (Supplementary Table 2 ). All sequences from the ship were collected between February 10 and 17, 2020. Hoshino et al. 8 The general time-reversible model with invariable sites and γ distribution among site rate variation (GTR+G+I) was selected as the best-fit model for the reference dataset using http://evomics.org/resources/software/molecular-evolution-software/figtree/) and were re-rooted using the earliest sample (IPBCAMS-WH-01, GenBank entry: MT019529.1). Bayesian phylogenetic analysis was conducted using a constant coalescent model and the strict molecular clock model implemented in BEAST (v.1.10.4) (Drummond and Rambaut, 2007) . The substitution rate of the SARS-CoV-2 sequences was calculated in a Bayesian (Drummond and Rambaut, 2007) , and ESS values >200 were accepted after burn-in. The best-fitting model was selected based on marginal likelihood estimation using path sampling and stepping-stone sampling methods (Baele et al., 2012) . Root-to-tip regression of the ML tree was analyzed using Tempest (v. 1.5.3). The root was determined according to the coefficient of determination, R 2 . The Bayesian skyline plot is a coalescent-based nonparametric method for estimating past population dynamics over time from molecular sequence data (Drummond et al., 2005) . The SARS-CoV-2 population dynamics on the Diamond Princess were inferred using a Bayesian skyline plot coalescent model with the strict molecular clock model implemented in BEAST (v.1.10.4) . The GTR+G+I substitution model was used, and the prior for the nucleotide substitution rate was set based on the values obtained in the rate estimation. The MCMC run contained 100 million states, with sampling every 10,000 states. We accepted ESS values >200 after a 10% burn-in, and three independent runs were combined using LogCombiner (v. 1.10.4) implemented in the BEAST package. The effective population sizes were plotted using R (v.4.0.2, R Core Team, 2018; https://www.R-project.org/). The R 0 is the predicted number of secondary cases that one case would generate in a completely susceptible population (Delamater et al., 2019) . Generally, an outbreak should continue if R 0 has a value >1, but not if R 0 is <1. The R 0 was estimated using the linear equation, R 0 = rD + 1, where r is the exponential growth rate and D is the average duration of infectiousness (Pybus et al., 2001; Wallinga and Lipsitch, 2007) . We calculated the exponential growth rate of 67 sequences derived from persons aboard the cruise ship using an exponential coalescent model with the strict molecular clock model implemented in BEAST2 (v.2.6.3). Three independent runs were combined. The average duration of infectiousness was calculated as 9.3 days (95% CI, 7.8-10) (He et al., 2020) , and the R0 was calculated using R (v.4.0.2)(Supplementary information). Doubling time was calculated as ln(2)/r. Limitations of this method include biases due to the duration of infectiousness and the estimated substitution rate (Pybus et al., 2001) . The birth-death skyline model is based on a forward-in-time model of transmission, death/recovery, and sampling (Stadler et al., 2013) . Temporal changes in R e during the outbreak were estimated using the birth-death skyline contemporary model implemented in BEAST2 (v.2.6.3). The GTR+G+I substitution and strict molecular clock models were used with nucleotide substitution rate estimates. For the birth-death model analysis, the epidemiological parameters comprising the rate of becoming noninfectious (δ), sampling proportion (ρ), R e , and the origin of the epidemic were established based on previous estimates (Stadler et al., 2013; Byrne et al., 2020; He et al., 2020; Lai et al., 2020a) . The rate of becoming noninfectious was calculated as the reciprocal of the average duration of infectiousness (9.3 days, 95%; CI 7.8-10) (He et al., 2020) , which corresponds to 39.5/year (95%CI: 36.5-46.8). Sampling proportion (ρ) is the proportion of lineages sampled in the analysis. The number of infected individuals up to February 17 was expected to range between 454 and 3711, which corresponds to the cumulative cases up to February 17, and the total number of individuals aboard the ship, respectively. Therefore, we calculated ρ to be between 0.018 and 0.14, and set a beta prior with parameters α = 2 and β = 15 (mean, 0.118; 95% CI, 0.0155-0.302). The R e of SARS-CoV-2 has been reported to be around 2-3 and 11 in the community-based infection and confined settings, respectively. A lognormal prior with a log-mean of 1 and an SD of 1.25 (mean, 5.94; 95% CI, 0.235-31.5) was used in this study. The origin of the epidemic is considered as the time point at which the index case became infected. We used 1/X prior as the origin of the epidemic prior. Three independent runs were combined as described previously herein, and R e was plotted using R (v.4.0.2) with bdskytools (https://github.com/laduplessis/bdskytools). The numbers of daily confirmed and cumulative cases obtained from the website of the Japanese Ministry of Health, Labour, and Welfare (https://www.mhlw.go.jp/index.html) and previous studies Yamahata and Shibata, 2020) were plotted using R (v.4.0.2) Hoshino et al. 13 An ML phylogenetic tree was constructed using the SARS-CoV-2 whole-genome Using the reference sequence dataset, the substitution rate was estimated under several settings (Supplementary Table 3 ). However, convergence was not achieved using the Bayesian skyline plot model. We selected strict clock and exponential coalescent models as providing the best fit based on marginal likelihood estimations. The The SARS-CoV-2 population dynamics on the Diamond Princess were inferred using the Bayesian skyline plot coalescent and strict molecular clock models. The Bayesian skyline plot showed that the effective population size began to increase around January 30 and exponentially increased between February 2 and 6, 2020 ( Figure 2a) . Subsequently, the population growth slowed down but continued on a slightly increasing trajectory, followed by a plateau phase from approximately February 10, 2020. The estimated mean tMRCA was 2020.066 (95% HPD, 2020.027-2020.095), corresponding to January 25, 2020 (95% HPD, January 10-February 4). We estimated epidemiological parameters and temporal changes in R e for the outbreak using the birth-death skyline contemporary model ( Princess cruise ship based on viral genome data using a Bayesian framework. Previous studies have estimated these transmission dynamics on the ship based on mathematical models using the daily number of confirmed cases or the daily number of symptomatic confirmed cases Nishiura, 2020; . However, these epidemiological data could have been biased because of the underestimation of infections owing to the limited diagnostic test capacity in the early phase of the outbreak and the limited sensitivity of RT-PCR (Woloshin et al., 2020) . In addition, the origin of the outbreak might be difficult to estimate due to the lack of diagnostic tests on the voyage. In contrast, recent advances in genomics and computing have facilitated the extraction of valuable information from viral genome data, and analytic methods have been developed to infer transmission history and past population dynamics of the pathogen with limited epidemiological data (Pybus et al., 2001; Drummond et al., 2005; Stadler et al., 2013) . We aimed to elucidate the dynamics of transmission from the perspective of genomic epidemiology, which would contribute to further understanding of the outbreak. SARS-CoV-2 can spread rapidly via close contact and in confined spaces , such as cruise ships (Moriarty et al., 2020; Sando et al., 2020; Sekizuka et al., 2020b) and airplanes (Hoehl et al., 2020; Pavli et al., 2020; Speake et al., 2020) . Our results showed the transmission dynamics of the SARS-CoV-2 outbreak in a high-transmission environment. The estimated origin of the epidemic was January 21, which coincided with the date when the presumed index case boarded the ship . The estimated R 0 was 2.56 (95% HPD, 1.04-4.65), which was in the range of the community-level R 0 (2.2-2.7) and in lower range of previous estimates for the ship (2.28-14.8) Zhang et al., 2020b) . The effective population size began to increase around January 30 and exponentially surged from February 2, before the start of quarantine on February 5. These findings indicate that most infections had already occurred before quarantine, which is consistent with previous findings ; National institute of infectious diseases, 2020; Nishiura et al., 2020) . When we compare our findings with epidemiological data, the exponential increase in confirmed cases after February 7 appears to follow the estimated substantial growth in infections by approximately 5 days. This time interval is in close agreement with the incubation period of the virus, which is on average 5.2 days with a 95th percentile of the distribution at 12.5 days . between the estimated increase in infections and the observed peak of daily febrile cases appears to be relatively close, considering the average incubation period of 5.2 (95% CI: 4.1-7.0) days . However, the incubation period of COVID-19 varies across individuals within a population, and factors such as age and disease severity affect the length of the incubation period (Dai et al., 2020; Kong, 2020; Lai et al., 2020b) . Therefore, the interval between our estimates and the number of febrile cases could be explained by the temporal change of infected subpopulations with different backgrounds. The modern cruise ships carry large numbers of people in limited spaces for a certain period, which facilitate the transmission of infectious diseases and outbreaks (Mitruka and Wheeler, 2008; Kak, 2015) . The Diamond Princess had many passengers and crew from various countries, and her cruise lasted more than 2 weeks. The facilities on the ship provided semi-closed spaces for passengers and crew, which contributed to high transmission of SARS-CoV-2. Large cocktail parties during the cruise played an important role in the dissemination of infections among passengers and crew (Yamagishi et al., 2020) . From the perspective of virology, SARS-CoV-2 has highly contagious characteristics, including a higher R 0 , longer incubation period, shorter interval between symptom onset and infectivity, and higher proportions of asymptomatic persons and patients with mild illness, compared to those with influenza, SARS-CoV, and MERS-CoV (Petersen et al., 2020) . These environmental and virologic factors might contribute to the outbreak of SARS-CoV-2 onboard the cruise ship. Hoshino et al. The R e peaked at 6.06 on February 4 and then decreased to 1.51. The effective population size continued to slightly increase after quarantine and reached a plateau around February 10. Although quarantine greatly reduced the number of infections, our results suggested that the infection spread slowly even after quarantine, which is consistent with previous epidemiological studies using mathematical models Nishiura, 2020) and haplotype network analysis using viral genome sequences (Sekizuka et al., 2020a) . Strict implementation of quarantine on the cruise ship was encumbered by difficulties in developing effective traffic lines or zones between infectious and noninfectious passengers and in providing daily support and medical procedures for passengers (Yamahata and Shibata, 2020) . The former possibly promoted transmission among passengers after quarantine through shared onboard facilities or cabins (Sekizuka et al., 2020a) . The latter necessitated activities of the crew, which put members at risk of infection. A previous study estimated that the epidemic peaked among passengers and crew between February 2-4 and February 8-10, respectively (Nishiura, 2020) . The epidemic period among the crew coincided with the increase in the effective population size after quarantine, suggesting that the slight suspected infected staff should be isolated as early as possible (Tokuda et al., 2020) . Finally, cooperation of various specialists and an established chain of command are essential for controlling a large-scale outbreak in confined settings (Tokuda et al., 2020) . Comprehensive and practical protocols should be designed for future epidemic events. Our study had certain advantages when compared with previous studies on the transmission dynamics of SARS-CoV-2 aboard the Diamond Princess using mathematical models and the number of laboratory-confirmed febrile cases Nishiura, 2020) . First, our study takes asymptomatic cases into consideration. Although previous studies focused on symptomatic population, a substantial proportion of SARS-CoV-2 infections is asymptomatic and has a significantly longer duration of viral shedding than that in the symptomatic population (Long et al., 2020; Yanes-Lane et al., 2020) , suggesting a critical role of asymptomatic individuals in the outbreak. In contrast, most sequences in this study were obtained from both symptomatic and asymptomatic patients through extensive tests, enabling our study to reflect the entire dynamics of transmission on the ship. Second, our analysis used objective data such as viral genome sequences and sampling dates. In contrast, the dates of fever onset can be affected by factors such as the background of individuals and testing situation on the ship. For instance, older adults with SARS-CoV-2 infection become more symptomatic and have a longer incubation period (Dai et al., 2020; Yamagishi et al., 2020) . In addition, distribution of thermometers among all individuals and the establishment of the Fever Call Center on February 7 could increase the detection of symptoms (Tsuboi et al., 2020) . The data Hoshino et al. 23 used in this study are more robust against these potential biases. Third, our study investigates the origin of the epidemic using viral genomic data. Our results showed that the outbreak had a single origin and that the epidemic started immediately after departure. However, in our study, one of the disadvantages was that the association between transmission dynamics and epidemiological factors could not be analyzed. Factors such as age, sex, location of cabin, and membership might largely affect transmission patterns. It was technically difficult to combine this information with phylogenetic analysis in this study. Therefore, we could not analyze the transmission dynamics in each subpopulation or between subpopulations. Second disadvantage is the potential sampling bias and sequencing errors due to next-generation sequencing techniques. It is difficult to obtain a complete and high-quality viral sequence from a sample with a low viral load. The viral load can be affected by the sampling procedure, immunological characteristics of individuals, and interval between infection and sampling. Therefore, there is a possibility that specific groups were excluded from our study. -This study has several potential limitations. First, we used SARS-CoV-2 sequences sampled over a period of 8 months. The short time range and relatively low diversity of SARS-CoV-2 might have affected the estimation of the substitution rate. To mitigate this challenge, we selected sequences to represent the temporal and geographical diversity of SARS-CoV-2 at the time while minimizing the impact of sampling bias. Epidemiological parameters of SARS-CoV-2 infection are currently limited. Although the prior used herein was set based on available epidemiological data, these settings might need modification based on future data. However, our results were consistent with those of previous epidemiological studies using mathematical models. Regarding the application of the Bayesian Skyline plot for epidemic data, sequences should be sampled sparsely from a large population under the assumption of a coalescent model. Sequences used in this study were less than 10% of the total confirmed cases, which might be sufficiently sparse. In addition, the HPD interval under a coalescent model tends to be overconfident when the number of lineages in the tree gets close to population size (Stadler et al., 2013) . Therefore, the HPD interval in the early phase of an epidemic should be interpreted with caution. R e was estimated using the birth-death skyline contemporary model in this study. The birth-death model can be biased largely when sampling process is mis-specified in the use of serially-sampled sequences (Volz and Frost, 2014) . It was difficult to establish a prior reflecting the complex sampling process on the ship, and convergence was not achieved. Because sequences were obtained almost contemporaneously, we used the birth-death contemporary model with the sampling proportion prior that can be set with reasonable certainty. This model selection could affect the estimated change of R e within approximately 1 day; however, our results would be robust against the irregular sampling process on the ship. Finally, we analyzed sequences as they were sampled from only one group. However, the transmission patterns differed between the passengers and crew, which might have affected our findings. The discrepancies in transmission patterns between passengers and crew can be explained by different living patterns on the cruise ship, locations of cabins in distinct decks, and extent of activity after quarantine (Nishiura, 2020; Tsuboi et al., 2020; Yamagishi et al., 2020) . The incorporation of the different transmission patterns into phylogenetic analyses would contribute to more accurate projections of transmission dynamics. However, it is difficult to incorporate phylogenetic analysis with complicated epidemiological models, including age, structure of host contact rates, or multiple host populations with different life history traits (Volz et al., 2013) . Alternatively, we could analyze sequences from a specific subpopulation, for instance passengers or crew. We identified four sequences from the crew in our dataset, which were too few to analyze transmission dynamics among the passengers and crew. Future studies involving more genetic and finer epidemiological data could facilitate further understanding of the complexity of SARS-CoV-2 transmission in confined spaces. In conclusion, we investigated the transmission dynamics associated with the COVID-19 outbreak aboard the Diamond Princess cruise ship. Based on the estimated substitution rate, phylodynamic analysis showed that the outbreak on the cruise ship occurred between late January and early February, before quarantine started on February 5, 2020. Although quarantine was clearly effective, the infection continued to slowly spread, even after the countermeasures were initiated. More effective measures should be considered to control outbreaks, particularly in confined settings. Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. None of the authors have any potential conflicts of interest to declare. -Infection spread substantially before quarantine on February 5 Improving the accuracy of demographic and molecular clock model comparison while accommodating phylogenetic uncertainty Presumed Asymptomatic Carrier Transmission of COVID-19 Novel cyclohexanone compound as a potential ligand against SARS-CoV-2 main-protease Inferred duration of infectious period of SARS-CoV-2: rapid scoping review and analysis of available evidence for asymptomatic and symptomatic COVID-19 cases Scientific Brief: SARS-CoV-2 and Potential Airborne Transmission The species Severe acute respiratory syndrome-related coronavirus: classifying 2019-nCoV and naming it SARS-CoV-2 Probable Longer Incubation Period for Elderly COVID-19 Cases: Analysis of 180 Contact Tracing Data in Hubei Province, China Complexity of the Basic Reproduction Number (R0) BEAST: Bayesian evolutionary analysis by sampling trees Bayesian coalescent inference of past population dynamics from molecular sequences Data, disease and diplomacy: GISAID's innovative contribution to global health Estimating the generation interval for coronavirus disease (COVID-19) based on symptom onset data Towards a genomics-informed, real-time, global pathogen surveillance system Nextstrain: real-time tracking of pathogen evolution Temporal dynamics in viral shedding and transmissibility of COVID-19 Assessment of SARS-CoV-2 Transmission on an International Flight and Among a Tourist Group Clinical features of patients infected with 2019 novel coronavirus in Wuhan Infections on Cruise Ships Longer incubation period of coronavirus disease 2019 (COVID-19) in older adults Tracking Changes in SARS-CoV-2 Spike: Evidence that D614G Increases Infectivity of the COVID-19 Virus Precision epidemiology for infectious disease control Early phylogenetic estimate of the effective reproduction number of SARS-CoV-2 Shorter incubation period is associated with severe disease progression in patients with COVID-19 Presumed COVID-19 Index Case on Diamond Princess cruise ship and Evacuees to Hong Kong Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia Using the contact network model and Metropolis-Hastings sampling to reconstruct the COVID-19 spread on the Clinical and immunological assessment of asymptomatic SARS-CoV-2 infections Genomic characterisation and epidemiology of 2019 novel coronavirus: implications for virus origins and receptor binding The situation report on the novel coronavirus infections Transmission potential of the novel coronavirus (COVID-19) onboard the diamond Princess Cruises Ship Estimating the asymptomatic proportion of coronavirus disease 2019 (COVID-19) cases on board the Diamond Princess cruise ship Public Health Responses to COVID-19 Outbreaks on Cruise Ships -Worldwide Backcalculating the Incidence of Infection with COVID-19 on the Diamond Princess Estimation of the asymptomatic ratio of novel coronavirus infections (COVID-19) In-flight transmission of COVID-19 on flights to Greece: An epidemiological analysis Comparing SARS-CoV-2 with SARS-CoV and influenza pandemics The epidemic behavior of the hepatitis C virus High population densities catalyse the spread of COVID-19 Diamond Princess cruise ship: estimating the epidemic potential and effectiveness of public health countermeasures Transmission of 2019-nCoV Infection from an Asymptomatic Contact in Germany COVID-19 outbreak on the Costa Atlantica cruise ship: use of a remote health monitoring system Haplotype networks of SARS-CoV-2 infections in the Diamond Princess cruise ship outbreak SARS-CoV-2 Genome Analysis of Japanese Travelers in Nile River Cruise COVID-19 vaccine development and a potential nanomaterial path forward Flight-Associated Transmission of Severe Acute Respiratory Syndrome Coronavirus 2 Corroborated by Whole-Genome Sequencing Birth-death skyline plot reveals temporal changes of epidemic spread in HIV and hepatitis C virus (HCV) COVID-19 outbreak on the Diamond Princess Cruise Ship in Epidemiology and quarantine measures during COVID-19 outbreak on the cruise ship Diamond Princess docked at Yokohama, Japan in 2020: a descriptive analysis Sampling through time and phylodynamic inference with coalescent and birth-death models Viral phylodynamics How generation intervals shape the relationship between growth rates and reproductive numbers Coronavirus disease (COVID-19) Weekly Epidemiological Up. Data as received by WHO from national authorities, as of WHO Director General's opening remarks at the media briefing on COVID-19-11 False Negative Tests for SARS-CoV-2 Infection -Challenges and Implications Nowcasting and forecasting the potential domestic and international spread of the 2019-nCoV outbreak originating in Wuhan, China: a modelling study Descriptive study of COVID-19 outbreak among passengers and crew on Diamond Princess cruise ship Preparation for Quarantine on the Cruise Ship Diamond Princess in Japan due to COVID-19 Proportion of asymptomatic infection among COVID-19 positive persons and their transmission potential: A systematic review and metaanalysis Crystal structure of SARS-CoV-2 main protease provides a basis for design of improved α-ketoamide inhibitors Estimation of the reproductive number of novel coronavirus (COVID-19) and the probable outbreak size on the Diamond Princess cruise ship: A data-driven analysis Branch length reflects units of nucleotide substitutions per site according to the provided scale Transmission dynamics of SARS-CoV-2 on board the Diamond Princess. (a) Effective population size of SARS-CoV-2 on board the Diamond Princess over time, (b) effective reproductive number (R e ) over time, and (c) daily and cumulative numbers of confirmed cases median estimate; colored area, 95% highest posterior density (HPD) confidence intervals of the estimate. Dashed (a) and dotted (b) lines, time to the most recent common ancestor and estimated time of outbreak origin, respectively. Abbreviations: COVID-19 SARS, severe acute respiratory syndrome; MERS, Middle East respiratory syndrome; RT-PCR, reverse-transcription PCR GTR, general time-reversible ESS, effective sample size HPD, highest posterior density tMRCA, time to the most recent common ancestor Masashi Mizokami: Conceptualization, Investigation, Supervision, Project administration, and Writing -review & editing Hoshino et al. We gratefully acknowledge the team at GISAID for creating the SARS-CoV-2 global database and the authors and investigators who submitted the sequences.