key: cord-0430086-t78lxy4z
authors: Francisco Junior, R. d. S.; Lamarca, A. P.; de Almeida, L. G. P.; Cavalcante, L.; Machado, D. T.; Martins, Y.; Brustolini, O.; Gerber, A. L.; Guimaraes, A. P. d. C.; Goncalves, R. B.; Goncalves, C. C. A.; Mariani, D.; Cruz, T. F.; de Souza, I. V.; de Carvalho, E. M.; Ribeiro, M. S.; Carvalho, S.; da Silva, F. D.; Garcia, M. H. d. O.; de Souza, L. M.; da Silva, C. G.; Ribeiro, C. L. P.; Cavalcanti, A. C.; de Mello, C. M. B.; Struchiner, C. J.; Tanuri, A.; de Vasconcelos, A. T. R.
title: Turnover of SARS-CoV-2 lineages shaped the pandemic and enabled the emergence of new variants in the state of Rio de Janeiro, Brazil
date: 2021-07-22
journal: nan
DOI: 10.1101/2021.07.20.21260890
sha: a34cb0935cb24e56e13e703582f518cdb35a3df4
doc_id: 430086
cord_uid: t78lxy4z

In the present study, we provide a retrospective genomic epidemiology analysis of the SARS-CoV-2 pandemic in the state of Rio de Janeiro, Brazil. We gathered publicly available data from GISAD and sequenced more 1,927 new genomes sampled periodically from March 2021 to June 2021 from 91 out of the 92 cities of the state. Our results showed that the pandemic was characterized by three different phases driven by a successive replacement of lineages. All stages occurred in distinct mortality and mobility contexts, with higher evidence of social distancing measures being observed in early pandemic and relaxed in the last two phases. Interestingly, we noticed that viral supercarriers accounted for the overwhelming majority of the circulating virus (> 90%) among symptomatic individuals in the state. Moreover, SARS-CoV-2 genomic surveillance also revealed the emergence and spread of two new variants (P.5 and P.1.2) firstly reported in this study. Altogether, our findings provided important lessons learned from the different epidemiological aspects of the SARS-CoV-2 dynamic in the state of Rio de Janeiro that have a strong potential to shape future decisions aiming to improve public health management and understanding mechanisms underlying virus dispersion.

One and a half years have passed since the outbreak of the Severe Acute Respiratory Syndrome-related Coronavirus 2 (SARS-CoV-2) virus in late 2019 in Wuhan, China. Currently, SARS-CoV-2 has accumulated a myriad of mutations and diversified into more than a thousand recognized lineages [1] .

While vaccination is advancing across the world and different treatments are being tested, the imminent emergence of resistant and more aggressive variants haunts the scientific community. In fact, lineages such as Gamma (P.1), Alpha (B.1.1.7), Beta (B.1.351), and Delta (B.1.617.2) are already known to partially evade natural-and vaccine-induced antibody response as well as have an elevated transmission rate [2] [3] [4] .

These Variants of Concern (VoC), named by the World Health Organization (WHO), have become widespread in the world replacing other lineages [5] [6] [7] .

The emergence of novel SARS-CoV-2 variants is particularly worrying in countries with slow vaccination rates, such as Brazil. Since March 2020, the COVID-19 pandemic has caused over 19 million cases and the loss of 530.000 lives so far. Although Brazil corresponds to approximately 2.7% of the total world population, it accounts for more than 12,5% of deaths globally. By July 2021, only 15% of the Brazilian population has completed the vaccination program recommended. This small number allows the virus to circulate freely within the population, enabling the emergence of mutations and further diversification into new lineages. As a matter of fact, previous studies analyzing the spatial progression of the pandemic in Brazil revealed that the states of São Paulo and Rio de Janeiro were pivotal to the dissemination of lineages throughout the country, including VoCs [8] [9] [10] [11] . Rio de Janeiro was also the background to the emergence and worldwide spread of the Variants of Interest (VoI) Zeta (P.2) [12] .

The high diversity of SARS-COV-2 lineages in the state of Rio de Janeiro since the beginning of the pandemic and the presence of three VoC and one VoI in the state is the perfect scenario to verify whether there is a relationship between lineage displacement and a surge in the number of cases. Here, we provide the first temporal analysis of the lineage dynamics in the state of Rio de Janeiro. By combining epidemiological and genomic data, we described the pandemic in three main phases in the state. In addition to genomes publicly available in GISAID, we have sequenced 1,927 new genomes from 91 out of the 92 cities in Rio de Janeiro, sampled periodically between March and June of 2021. Finally, we report the emergence of two new variants, P.1.2 and P.5, with the reconstruction of the within-country and worldwide dispersion.

Samples from patients with SARS-CoV-2 positive nasopharyngeal RT-PCR were collected at the Noel Nutels Central Laboratory (LACEN-RJ) and Unidades de Apoio ao Diagnóstico da (UNADIG-RJ). The 

Epidemiological data were retrieved from the Centro de Informações Estratégicas e Resposta de Vigilância em Saúde (CIEVS-RJ) of the Secretaria de Saúde do Estado do Rio de Janeiro. We then estimate the time varying reproduction numbers (Rt) from the Epidemic Curve using the EpiEstim package in R [13] . To further characterize the mobility patterns during the pandemic, we downloaded mobility data for the cities in the state of Rio de Janeiro from Google Community Mobility Reports (available at: https://www.google.com/covid19/mobility/) between 1 March 2020 to 6 June 2021. Six indices for variation in the expected value of the time typically spent in residential, transit stations, parks, groceries and pharmacies, retail and recreational, and workspaces were summarized using the 7-day moving average considering the whole state.

To verify whether a small group of samples can be overrepresented the Ct values in our dataset. We selected 1,119 out of the 1,933 samples sequenced and retrieved the cycle threshold (Ct) values of the target virus genes and the endogenous gene RnaseP. We performed a relative quantification of the viral loads (RQVL) using the method of 2 -∆CT . The dataset table with sample ID, sex, age, and RQVL colunms was descend ordered by the RQLV values. We calculate the ratio of 2%, 10%, and 20% of the samples relative to the total sum of the RQVL. We also performed a generalized linear model using the gamma distribution with . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted July 22, 2021. ;  the inverse link function to verify the significance of sex and age in the RQLV distribution using stats package in R [14] .

In order to infer their phylogenetic tree, we have retrieved all genomes from the state of Rio de Janeiro available in GISAID by June 21 st ,2021 that were longer than 29,000 bases, had high coverage (as defined by GISAID) and had complete collection dates (n = 3,029). We used MAFFT 7.475 with --auto and --addfragments options [15, 16] to align this dataset to the WH01 (EPI_ISL_406798) genome sampled in 2019 from Wuhan, China. The 3' and 5' untranslated regions were then trimmed. The subsequent alignment was used to construct the maximum likelihood tree using IQ-TREE 2 [17] with the GTR+F+I+G4 substitution model, selected with the ModelFinder algorithm [18] . Assuming a constant evolutionary rate of 8 x 10 -4 substitutions/site/year, we rescaled the branch lengths to dates with TreeTime [19] . Tips that deviated from the clocklike model are ignored by TreeTime ( Figure S1A ) and were removed from the final tree.

To reconstruct the evolution of the new P.1.2 lineage described in this work, we have retrieved all sequences from this lineage available in GISAID by June 21 st and aligned them to our newly-sequenced samples. As an outgroup, we selected the two oldest P.1 sequences available from the state of Amazonas. We then reconstructed the maximum likelihood tree with IQ-TREE 2 and the GTR+F+I model selected with

ModelFinder algorithm. We then tested the sequences' conformity to the molecular clock model by analyzing the correlation between root-to-tip distances and tip dates and removed from the alignment all sequences that were not included within the 95% confidence interval of the regression ( Figure S1B ). As the two outgroup sequences selected were removed from the final dataset, we selected as new outgroups two sequences from the P.1 clade positioned closest to P.1.2 according to the maximum likelihood phylogeny of all sequences from Rio de Janeiro.

Divergence dates were estimated using a Bayesian approach with BEAST 1.10.4 [20] . Models selected for the analyses were the GTR+F+I with estimated base frequencies, the GMRF Skyride population model, and the strict clock model. The clock rate prior was defined as a uniform distribution with 6E-4 and 1E-3 as minimum and maximum values. The MCMC was run through 200,000,000 steps with sampling every 10,000th and a burnin of 10%. Lowest ESS was 204, from the "treeLength" parameter. In order to evaluate the impact of biased sampling through time on population inference, we also applied the Bayesian Nonparametric Phylodynamic Reconstruction model with and without preferential sampling correction on the consensus tree of P.1.2 using the R package phylodyn [21] .

The consensus tree was used as fixed topology and branch lengths to reconstruct the locality of its ancestor nodes. We used the Symmetric trait substitution model with the Bayesian Stochastic Search Variable . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 22, 2021. ; https://doi.org/10.1101/2021.07.20.21260890 doi: medRxiv preprint Selection approach for rate estimation in BEAST 1.10.4. The MCMC chain was 10,000,000 steps with sampling every 10,000th and a burnin of 10%. Lowest ESS value was 166, from the "loc.indicators.Argentina.Paraguay" parameter. Ancestor locations inferred were then converted into coordinates and mapped using the SERAPHIM package [22] in R software.

We also reconstructed the evolutionary tree of the new lineage P.5. To accomplish this we identified all the sequences available in the GISAID database by June 21 st that carried the mutations that characterize the lineage. We aligned these sequences to the WH01 genome using MAFFT 7.475 with --auto and --addfragments options and trimmed the 3' and 5' untranslated regions. The maximum likelihood tree was then inferred using IQTREE with the GTR+F+I model, selected by the ModelFinder algorithm.

The primary sequences of Spike protein from P.1 lineage and P.1.2 subclade (A262S) were submitted in the Swiss-Model web server (https://swissmodel.expasy.org/) to obtain the best templates and build their 3D structures. The stability analysis was performed using the Stability function from FoldX tool [23] . Salt bridges were measured using the EBRIS web server [24] . The docking analysis was performed between structures of P.1 and P.1.2 subclade in HDOCK [25] . We evaluated the Spike protein against ACE2 protein and four types of antibodies anti-Spike extracted from CoV-AbDab database [26] . The human antibodies were selected based on their preferences for distinct binding locations in Spike: subunits S1, S2, S1-S2, S1-RBD, and S1-notRBD. The ACE2 structure was extracted from a complex reported in PDB with RBD region. We compared the five complexes generated for each protein using the residues found in contact with ligand and belonging to the hotspot predicted area. In the first part of the docking analysis, we compared the amino acids contacts between the receptor (P.1/P.1.2) and the five ligands. The binding affinity of each complex was measured with the Prodigy tool [27] . We searched for hot spots in the receptors using the KFC (Knowledge-based Fast atomic density evaluation and Contacts) webserverer [28] . The potential profiles considering P.1.2 subclade structure and P.1 were compared by PDB2QR [29] and APBS. Visualization and structure manipulation was conducted using PyMol.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 22, 2021.

In early March 2020, Rio de Janeiro's Secretary of Health confirmed the first case of COVID-19 in the state. Since then, the SARS-CoV-2 virus has rapidly spread causing more than 923,234 cases and 54,125 deaths in the state of Rio de Janeiro only, with multiple events of the public health system collapsing. The highest prevalence of COVID-19 cases was observed among the central and north regions in the state (Table   S1 ). We also noticed a high number of cases in the metropolitan region, known to have high population density, workers commuting daily between neighboring cities and an elevated number of interstate corporate travels. A statistically significant difference was found across the number of cases and deaths between females and males (OR 0.73, 95% CI: 0.71-0.74; p-value<2.2×10 −16 ; Figure S2A ). Though more females got infected, male mortality was higher. Among confirmed cases, typically younger subjects ranging between 30-49 years old (y.o) represented the highest occurrence rate. Nevertheless, the percentages of mortality surpassed 11% in patients aged >60 y.o reaching more than 79% in the oldest group ( Figure S2B ).

The epidemiological findings suggest three different exponentially growing phases associated with the disease in the state ( Figure 1A ). Firstly, after the epidemic outbreak in March 2020, the number of cases increased, reaching a peak in May 2020 when it started to decline. In early November 2020, a period The pairwise comparison between the mortality rate across the different phases revealed substantial differences among the first and second stages (p-value: 2.2×10 −4 , Wilcoxon test -Benjamini-Hochberg p-value correction), the second and third (p-value: 3.2×10 −05 ) and the first and third phases (p-value: 7.9×10 −10 ) ( Figure S2C ). Differential mortality rate was also identified among the age groups ( Figure S3 , Table S2 ), the most recent phase seemed to be more lethal among subjects ranging between 20 to 69 y.o than the others. On the other hand, the mortality in the second phase was consistently decreased among young ages and increased in subjects older than 79 y.o ( Figure S3 ).

Remarkably, mobility-driven changes significantly accounted for the virus dispersion in the state ( Figure 1C and D). The early phase was mainly characterized by a sharp decrease in activities outside home such as . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 22, 2021. ; https://doi.org/10.1101/2021.07.20.21260890 doi: medRxiv preprint time spent in workplaces (-64%), retail and recreation (-74%), official national park visits (-72%), transit station (-67%), and going to grocery and pharmacy (-40%) in late March 2020. In the same period, we also observed an increase in the expected value of the typical day (baseline) in the time spent at home (+25%), evidencing the social distance measures in Rio de Janeiro. These numbers rose close to the baseline in workplace activities and decreased in households during the second and third phases as the social distancing measures were relaxed. We also noticed a continuous elevation in the percentage of going to grocery and pharmacy from -40% (1 st phase) to +74% (3 rd phase) over time, suggesting that people moved to these places more often than the baseline frequency. Therefore, both phases occurred in distinct mobility contexts with the most prominent change being reported in the first phase followed by similar patterns in the other two.

After the introductory events in early March 2020, independent efforts to sequence SARS-CoV-2 genomes from Rio de Janeiro resulted in 3,932 sequences publicly available in the GISAID database until June 2021.

This number corresponded to an approximate overall average of four sequences per 1000 cases, reaching a peak of 1% of all reported cases sequenced in April and May 2021 (Table S3) . We observed that the pandemic phases were characterized by successive replacement of SARS-CoV-2 lineages in the state ( Figure 1E ). The first phase was predominantly driven by the lineage B.1.1.33, which was the most prevalent strain circulating during eight months, from March to October 2020, being gradually replaced by P.2 during the second phase. In May 2020, B.1.1.33 achieved the higher relative frequency of approximately 92% of the genomes deposited in GISAID, coinciding with the peak reported during the first phase.

Moreover, the dominance of P.2, from November 2020 to January 2021, was preceded by a continuous increase in the frequency of its ancestor, B.1.1.28, that was already circulating in a small frequency in the state since April 2020. The peak of reported cases observed in the second phase matches the highest prevalence of P.2 in December 2020. After this, P.1 became the dominant lineage reaching a frequency greater than 90% from March 2021, matching the peak of cases in the third phase. Therefore, the introduction and emergence of each lineage occurred concomitantly with the increase in the number of cases.

From the 1,927 new genomes sequenced in this study after later March 2021, 92.42% were assigned to P.1 (Gamma) lineage. We then sought to investigate the distribution of SARS-CoV-2 viral loads within a population of 1,119 samples randomly selected from the state. The cumulative distribution of the relative quantification of the viral loads (RQVL) showed that 2% (n = 22, group A) of the individuals carried 60% of the circulating virions in the period analyzed (Figure 2A ). This value increases to 92.4% of the virions harbored by just 10% (n = 112, group B) of the population (Figure 2A) . Furthermore, 20% (n = 223, group C) of our samples represented 98.2% of the viral loads of our dataset (Figure 2A ). Next, we tested the contribution of age, sex and the interaction between age and sex to the viral loads using a gamma-generalized linear model (GLM). Both groups (A, B, and C) had approximately half of samples . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 22, 2021. ; https://doi.org/10.1101/2021.07.20.21260890 doi: medRxiv preprint from each sex, nevertheless, a higher relative viral load were reported among males within-group (Figure 2 B-D). From the 22 samples (11 males and 11 females) present in group A, the viral load was 14.8% greater in males when compared to females. Among males in group B and C, our model showed that subjects aged > 40 y.o carried most of the population-circulating virions, and the older the higher the viral load observed (Table S4 ; Figure S4 ). No statistical differences were observed among females age groups.

SARS-CoV-2 genomic surveillance in the state of Rio de Janeiro also allowed the identification of two new variants (P.5 and P.1.2) first reported in this study ( Figure 3A) . compared to the template, stability analysis showed that P.1.2 is less stable than P.1 according to FoldX (Table S5) . This difference was confirmed by estimating the number of salt bridges in both models, P.1 had more salt bridges (n = 53) than P.1.2 (n = 48), which demonstrates that P.1 folding is less flexible. We also investigated possible effects in the electrostatic surface caused by the change of an Alanine to Serine in position 262. Figure S7A and B shows that both proteins present the same potential surface although in the . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 22, 2021. ; https://doi.org/10.1101/2021.07.20.21260890 doi: medRxiv preprint mutated region the cavity becomes more neutral in P.1.2, possibly altering the binding-site of the region.

Interaction between both models with four antibodies and ACE2 surface protein using docking and affinity binding prediction (Table S6 ; Figure S7 ) showed that P.1.2 was more efficient in antibody interaction than P.1. P.1 also performs more stable bonds with ACE2 (-18.2 kcal/mol) than P.1.2 (-17.8 kcal/mol) and is predicted to enter easier in the cell to establish virus infection. Furthermore, we detected an active binding site in the complex S1notRBD-P.1.2 that is a candidate to alter the binding site contacts in the cavity ( Figure   S7 ).

In this study, we gathered epidemiological and genomic data from the outbreak of SARS-CoV-2 until June Whereas the daily number of cases increased during the second and third phase, more people died in the first and third ones. We hypothesized that lack of knowledge about the disease and resources to treat patients in early pandemic may explain such severity in the first phase [30] [31] [32] . After almost a year of pandemic, a significant amount of information had been already gathered about the virus and the disease, certainly improving patient care. We have observed a switch in the death profiles over the course of pandemic, while deaths were more restricted to old groups and people with comorbidities in the first phase, a substantial increase of young people was reported at the third wave. Prioritizing the vaccination of elderly people might have had a strong impact on reducing the mortality of this group [33] , though it does not explain the rise in mortality in younger ages. Thus, it is plausible to assume that the evolution of new viral phenotypes are correlated to the rejuvenation in mortality observed in the third stage [34] [35] [36] [37] .

The displacement of SARS-CoV-2 lineage across each phase suggests that the new successive lineage had higher adaptive value than the last one [4, 6] . Indeed, the three lineages progressively accumulated aggressive key mutations associated with higher transmissibility or evasion to the host immune system response: namely, B.1.1.33 carried D614G, P.2 harbors D614G and E484K, and P.1 carries N501Y in . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 22, 2021. addition to the two previous mentioned. The observed synchronicity between lineage substitution and a rise in the number of cases may be a reflection of this increase in adaptability of the virus to human biology. This dynamic raises a red flag to the aggressiveness of the next dominant lineage. Though lineage displacement is a well-documented aspect of pathogen evolution [38] [39] [40] , the mechanisms and models underlying the influence of this event on the resurgence of local outbreaks of COVID-19 is still poorly understood. Nevertheless, displacement of lineages by B. 1.1.7, B.1.351, B. 1.617.2 and P.1 has been reported repeatedly across the world [5, 7, [41] [42] [43] [44] .

We caution that direct association between each phase and the lineage distribution must be interpreted considering the temporal bias in genome sequencing, especially during the first phase (March-October 2020). Nevertheless, previous studies indicate that B.1.1.33 was widely disseminated throughout the country in the early pandemic phase [8, 9] , supporting our findings for Rio de Janeiro. Furthermore, the concentration of samples in the metropolitan region challenges the exact dispersion routes within-state as also highlighted by Wilkinson and colleagues [45] . Our results demonstrated that returning to work-related and outside-home activities close to the pre-pandemic levels preceded the rise in transmission rates, especially in the second phase. The surge in cases, therefore, could also have been triggered by the gradual flexibilization of social distancing measures [46] . As the number of cases never shrank to pre-pandemic levels in either phase, the slight increase in transmission rates caused by this flexibilization would be enough to create the exponential rise in the virus population. Though the "lineage" and the "mobility" hypotheses could be perceived as opposing narratives, we reiterate that both endogenous and exogenous factors control the threshold of a new surge in cases. The introduction of more aggressive lineages requires the enforcement of harder non-pharmacological interventions to control the local outbreak. So far, the only way to circumvent this relationship is the reduction in viral transmission rates through vaccination. Further epidemiological modelling is still needed to better evaluate the influence of all mechanisms involved in this dynamic.

An important aspect of the disease dynamic revealed by this study was that viral supercarriers accounted for the overwhelming majority of the circulating virus in the state. Recent studies demonstrated that transmission of COVID-19 is more likely to occur in individuals with higher viral load [47] [48] [49] . Given that most of the samples in the period analyzed were assigned to P.1, such a mechanism could also shed light on the rapid spread of this VoC. Considering that the selection of RT-PCR positive samples for sequencing in our analysis is based on a pre-filtering of samples with Ct values < 27, the values of 60% and >90% of viral load of the supercarries could represent an even bigger amount of the circulating virus in the population.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted July 22, 2021. ; https://doi.org/10.1101/2021.07.20.21260890 doi: medRxiv preprint Indeed, Yang et al recently described a bigger fraction of viral load (90%) being carried by just 2% of the infected individuals [50] , which corroborates our findings. Though the observed viral load in our cohort could be explained by the different stages of infection of each patient, it is known that this measure is significantly variable between individuals [51, 52] . Overall, supercarriers may play an important role in the dynamic of virus dissemination in the state.

It is well known that increasing viral replication rate accelerates the emergence of new mutations and, therefore, the diversification into new lineages. It is not surprising that the origin of P.1.2 is simultaneous to the countrywide propagation of P.1 in early January 2021 [10, 53, 54] , when this rate is expected to be high.

The new mutations observed in P.1.2, though, do not indicate that this lineage has a higher adaptive value than P.1. Indeed, structural protein modelling of its sub-clade suggests a higher instability and better docking with antibodies than P.1. The fast dispersal of P.1.2 within Brazil could be attributed to an opportunistic event. As P.1 was introduced in Brazilian states at the end of 2020, it induced an immune response in the host population, which temporarily prevents reinfection and limits the number of available hosts. Therefore, the P.1.2 population could only grow before P.1 had not achieved such high frequency, as observed after January [39] . Same behaviour could explain the emergence and spread across the state observed by the lineage P.5. A new opportunity of growth opens after infection-induced immune response fades in the host population, which can only be prevented through vaccination.

Finally, learning from past efforts to control the pandemic shapes the future decisions towards an improvement of the national public health, especially when embracing the findings reported by the scientific community. Currently, as this manuscript is being written, the state is at imminent risk of experiencing a fourth phase of COVID-19 epidemic due to the introduction of VoC delta (B.1.617.2) as recently reported by us [55] . Therefore, understanding the different epidemiological aspects controlling the SARS-CoV-2 dissemination, such as host mobility patterns and viral evolution, is essential to elaborate effective intervention measures that reduce the number of affected individuals. These evaluations are particularly vital in regions with high-density populations, which are prone to spread viruses worldwide.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Community Mobility data reports with changes in time spent in workplaces, residential, retail and recreation, grocery and pharmacy, parks, and transit stations. E) Temporal distribution of the relative frequencies of SARS-CoV-2 lineages in the state of Rio de Janeiro.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Table S4 .

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted July 22, 2021. ; https://doi.org/10.1101/2021.07.20.21260890 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted July 22, 2021. ; https://doi.org/10.1101/2021.07.20.21260890 doi: medRxiv preprint

Addendum: A Dynamic Nomenclature Proposal for SARS-CoV-2 Lineages to Assist Genomic Epidemiology

Reduced Neutralisation of the Delta (B.1.617.2) SARS-CoV-2 Variant of Concern Following Vaccination

INCA COVID-19 Task ForceˆNew Infections by SARS-CoV-2 Variants of Concern after Natural Infections and Post-Vaccination in Rio de Janeiro

Multiple SARS-CoV-2 Variants Escape Neutralization by Vaccine-Induced Humoral Immunity

A Global Analysis of Replacement of Genetic Variants of SARS-CoV-2 in Association with Containment Capacity and Changes in Disease Severity

Detection of a SARS-CoV-2 Variant of Concern in South Africa

Evolutionary Dynamics and Dissemination Pattern of the SARS-CoV-2 Lineage B.1.1.33 During the Early Pandemic Phase in Brazil

Evolution and Epidemic Spread of SARS-CoV-2 in Brazil

Genomic Surveillance of SARS-CoV-2 Tracks Early Interstate Transmission of P.1 Lineage and Diversification within P.2 Clade in Brazil

Epidemiological Dynamics of SARS-CoV-2 VOC Gamma in Rio de Janeiro, Brazil

Genomic Characterization of a Novel SARS-CoV-2 Lineage from Rio de Janeiro, Brazil

A New Framework and Software to Estimate Time-Varying Reproduction Numbers During Epidemics

R Core Team R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing

MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability

Adding Unaligned Sequences into an Existing Alignment Using MAFFT and LAST

IQ-TREE 2: New Models and Efficient Methods for Phylogenetic Inference in the Genomic Era

Fast Model Selection for Accurate Phylogenetic Estimates

Maximum-Likelihood Phylodynamic Analysis

BEAST: Bayesian Evolutionary Analysis by Sampling Trees

Phylodyn: An R Package for Phylodynamic Simulation and Inference

SERAPHIM: Studying Environmental Rasters and Phylogenetically Informed Movements

FoldX as Protein Engineering Tool: Better Than Random Based Approaches?

Statistical Characterization of Salt Bridges in Proteins

The HDOCK Server for Integrated Protein-protein Docking

The Coronavirus Antibody Database

PRODIGY: A Web Server for Predicting the Binding Affinity of Protein-Protein Complexes

Interactive Forecasting of Protein Interaction Hot Spots

PDB2PQR: An Automated Pipeline for the Setup of Poisson-Boltzmann Electrostatics Calculations

COVID-BioB Study Group Decreased in-Hospital Mortality in Patients with COVID-19 Pneumonia

Trends in Covid-19 Risk-Adjusted Mortality Rates in a Single Health System

Lower Mortality of COVID-19 by Early Recognition and Intervention: Experience from Jiangsu Province

Estimating the Early Impact of Vaccination against COVID-19 on Deaths among Elderly People in Brazil: Analyses of Routinely-Collected Data on Vaccine Coverage and Mortality

Second Wave of COVID-19 in Brazil: Younger at Higher Risk

Competition and the Origins of Novelty: Experimental Evolution of Niche-Width Expansion in a Virus

The Emergence of Novel SARS-CoV-2 Variant P.1 in Amazonas (Brazil) Was Temporally Associated with a Change in the

Sudden Rise in COVID-19 Case Fatality among Young and Middle-Aged Adults in the South of Brazil after Identification of the Novel B.1.1.28.1 (P.1) SARS-CoV-2 Strain: Analysis of Data from the State of Parana

Superior Infectivity for Mosquito Vectors Contributes to Competitive Displacement among Strains of Dengue Virus

Modelling Pathogen Competition and Displacement-Phytophthora Infestans in Scandinavia

Long Intervals of Stasis Punctuated by Bursts of Positive Selection in the Seasonal Evolution of Influenza A Virus

Increased Transmissibility and Global Spread of SARS-CoV-2 Variants of Concern as at

Emergence and Spread of SARS-CoV-2 Lineages B.1.1.7 and P.1 in Italy

Molecular Analysis of SARS-CoV-2 Genetic Lineages in Jordan: Tracking the Introduction and Spread of COVID-19 UK Variant of Concern at a Country Level

A Year of Genomic Surveillance Reveals How the SARS-CoV-2 Pandemic Unfolded in Africa

An Optimal Predictive Control Strategy for COVID-19 (SARS-CoV-2) Social Distancing Policies in Brazil

Temporal Dynamics in Viral Shedding and Transmissibility of COVID-19

Transmission of COVID-19 in 282 Clusters in Catalonia, Spain: A Cohort Study

Higher Viral Load Drives Infrequent SARS-CoV-2 Transmission between Asymptomatic Residence Hall Roommates

Just 2% of SARS-CoV-2−positive Individuals Carry 90% of the Virus Circulating in Communities

Viral Dynamics of Acute SARS-CoV-2 Infection

SARS-CoV-2 Viral Load in Upper Respiratory Specimens of Infected Patients

SARS-CoV-2 Lineage in Manaus

COVID-19 in Amazonas, Brazil, Was Driven by the Persistence of Endemic Lineages and P.1 Emergence

Genomic Surveillance of SARS-CoV-2 in the State of Rio de Janeiro, Brazil: Technical Briefing Available online

We would like to thank all the authors and administrators of the GISAID database, which allowed this study of genomic epidemiology to be conducted properly. A full list acknowledging the authors publishing data used in this study can be found in the Supplementary File. We are very grateful to Dr. Luiz Max Fagundes de Carvalho from Fundação Getúlio Vargas for the kind aid with Bayesian models and analyses.