key: cord-0683907-905u3mq5
authors: Lu, Y.; Han, K.; Xue, G.; Zheng, N.; Jin, G.
title: Estimated Spike Evolution and Impact of Emerging SARS-CoV-2 Variants
date: 2021-05-10
journal: nan
DOI: 10.1101/2021.05.06.21256705
sha: 4318d5662b4ad281fb75a65afd22238b03893182
doc_id: 683907
cord_uid: 905u3mq5

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that causes COVID-19, has been mutating and thus variants emerged. This suggests that SARS-CoV-2 could mutate at an unsteady pace. Supportive evidence comes from the accelerated evolution which was revealed by tracking mutation rates of the genomic location of Spike protein. This process is sponsored by a small portion of the virus population but not the largest viral clades. Moreover, it generally took one to six months for current variants that caused peaks of COVID-19 cases and deaths to survive selection pressure. Based on this statistic result and the above speedy Spike evolution, another upcoming peak would come around July 2021 and disastrously attack Africa, Asia, Europe, and North America. This is the prediction generated by a mathematical model on evolutionary spread. The reliability of this model and future trends out of it comes from the comprehensive consideration of factors mainly including mutation rate, selection course, and spreading speed. Notably, if the prophecy is true, then the new wave will be the first determined by accelerated Spike evolution.

According to the statistics of World Health Organization (https://covid19.who.int/), pandemic has caused at least 131 million confirmed cases and more than 2.8 million deaths as of April 4, 2021 . During global circulating, the SARS-CoV-2 has been mutating despite at a slower rate than influenza and human immunodeficiency virus (HIV) (1-5). Not surprisingly, a notorious variant named B.1.1.7 has infected more people than other strains (6) . Its occurrence not only raised new questions on the higher infectivity and fatality, but also suggested uneven evolution paces on variants' mutation and selection, separately and uniquely. Thus, new bioinformatics tools are in urgent demand to develop to answer and confirm if a dynamic evolution could be true by tracking mutation rate.

To achieve this goal, efforts have been devoted to studying viral phylodynamic by finding and analyzing clades (3) . GISAID (https://www.gisaid.org) maintains complete SARS-CoV-2 genomes and had more than 800,000 genomes in the database as of March 2021. The phylogenetic analysis has identified 9 clades in GISAID which are S, L, O, V, G, GH, GV, GR, and GRY. Among them, clades S and O were the most common ones in January and February 2020, whereas clades G and GRY prevailed from December 2020 to March 2021, especially the latter one. Given the widespread appearance and large numbers of B.1.1.7 genomes globally surpassing numbers of other existing clades, GISAID elevated the B.1.1.7 lineage to clade GRY that was descended from clade GR.

More importantly, these data have facilitated the mutation surveillance on Spike protein to understand how the pandemic shifts characteristics over time. A major form of the Spike 3 mutations is D614G (3, (7) (8) (9) (10) . It reduces S1 shedding and increases infectivity. Also, it is the marker mutation for clades G, GH, GV, GR, and GRY. Unexpectedly, at the genomic level, B.1.1.7 not only accumulated mutations that failed nucleic acid testing (11) , but have more nucleotide changes on the genomic region of the C-terminal domain of Spike protein with yet unknown importance (12) . Accordingly, took nonsense mutation together with all other types into consideration, we developed a bioinformatics model to study (1) the speed of Spike genomic evolution and further (2) how long it would take by a new variant to ignite another outburst. Based on and supported by this model, we found that most emerging variants were generated from a small portion of relatively original mutated strains driven by accelerated evolution. Furthermore, newly emerged variants cannot immediately cause a pandemic peak but need to show highly selective advantages within one to six months among a large population. These findings could benefit a scientific understanding of the dynamic virus evolution and benefit practical predicting of an upcoming wave.

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint

Our bioinformatics approach recorded mutation numbers of mismatches and indels in Spike coding region by first aligning GISAID-derived 817,351 SARS-CoV-2 sequences as of March 15 th 2020 to the WIV04-Spike reference (13) , using the BLAST software (5) (Methods), and then normalized them to the length of the Spike genomic region. Next, Mutated Number per Kilobase (MNPK) was calculated by multiplying the normalization results by 1,000 and defined as Spike mutation rate (Fig. S1) .

Further analysis showed that in gradually disappearing clades (L, O, S, and V), high MNPKs densely happened in early time points but later disappeared or dramatically decreased to a lowfrequency level where dots were way sparse and values below 5, most around 2.5 (Fig. 1) . While in currently prevalent clades (G, GH, GV, and GR), oppositely, high MNPKs clearly increased to a serious enrichment in at least one continent (Fig. 1) . This suggested that these Spike coding regions are still swiftly mutating without an obvious peak thus new variants could continuously appear. Furthermore, steady highest MNPKs only showed at most recent time points in the newest clade GRY (14) , especially in Europe and North America (Fig. 1) . Lastly, SARS-CoV-2 genomes not assigned (NA) into any clades defined by GISAID also showed some mutations outnumbered by GRY but still accumulated in 3 continents (Fig. 1) . A question that if another new variant would come out of them doesn't have an answer yet but need long-term monitoring.

Taken together, our bioinformatic approach detected a trend of Spike mutation rates which is in accordance with the relative chronological order of each clade's occurrence yet not necessarily associated with the population size of continents. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Since the trend we observed is quite similar to reported significantly increased mutation rates on emerging variants such as B.1.1.7 (8, 9, 12, 14) , we then reasonably hypothesized that the differences in Spike genomic mutation rates could reflect a revolutionary strategy in which its pace was elevated for SARS-CoV-2 to generate more infectious variants, no matter from single mutation or recombination (15), and dropped after task completion then precedent variants disappeared. This process could be a result of accelerated evolution or viral recombination. If this is true, then Spike genomic region evolves at different speeds in different variants.

To verify the hypothesis, we defined Relative Evolutionary Rate of Spike (RERS) by MNPK over time (Methods) and measured it according to (i) to each clade in a continent, SARS-CoV-2 sequences were grouped every seven sampling days from January 2 nd 2020 to March 15 th 2021 and arithmetic mean MNPKs were calculated; (ii) RERS i (i = 2, 3, 4, …, g) was calculated by the following equation.

, positive and negative values mean increased and decreased Spike evolution, respectively.

Through further analysis at the clade and geographic-region level, it showed that to gradually disappearing clades (L, O, S, and V),

values either cannot be observed or All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint sparse, suggesting that Spike genomic region in these four clades stopped or hugely slowed down their revolution, consistent with the fact that these clades are gradually disappearing from the pandemic. On the contrary, an obvious acceleration can be observed in currently prevalent clades G, GH, GV, and GR (Fig. 2) . Besides, in each continent, at least one clade is evolving at an elevated speed, in Europe even all four. This clearly suggested that these clades are still globally dominant and undergoing speed-up evolution everywhere.

Next,

in clade GRY hasn't increased since February 2021 ( Fig. 2) though it has the highest MNPK ( Fig. 1) . This may be attributed to the huge selective pressure brought about by vaccines, together with the fact that not enough time nor input sequences specifically for this clade. So, the evolutionary situation of GRY should still be monitored.

Finally, one increased and one decreased RERS were observed from NA SARS-CoV-2 sequences in Europe and North America ( Fig. 2) with currently unclear influence.

Conclusively,

distribution showed that faster Spike evolution is driven by accelerated mutation in its genomic region thus governs the ongoing pandemic and/or potentially generates new variants possibly responsible for next wave, and thus confirmed our hypothesis.

We further explore more potentials of RERS in a more general manner. Similar to

was calculated by all SARS-CoV-2 sequences but without restrictions from clades and continents.

From ܴ ‫ܧ‬ ܴ ܵ ீ fitting curve, we can clearly see a three-staged, two-turning-point pattern ( Fig.   3A ): It initially vigorously rocketed from January 13 th 2020 to the peak on May 29 th 2020, then All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. To provide an answer, we turned back to compare MNPK of all clades at each selected time point. Quite differed from distribution results, clade L, but not the largest V, on February 19 th 2020 showed overwhelmingly high MNPK over the rest (Fig. 3B) ,

suggesting that it was contributing the most mutations to the first rising phase, at least at that point. This characteristic ended on May 29 th 2020 when mutation and revolution all reached the peak (Fig. 3B) . On October 26 th 2020 when the wave was about to start, most SARS-CoV-2

Spike mutations were generated from clade GRY. In the second climbing phase, clades O and GR had the most mutations (Fig. 3B) . These data together suggested that during a rapid All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. peak on May 29 th 2020 to that of the pandemic peak on November 28 th 2020 (Fig. 3A) . We then compared new_cases, new_deaths, new_cases_per_million_people, and new_deaths_per_million_people of the two days to gain more insights into the pandemic and found that numbers of ܴ ‫ܧ‬ ܴ ܵ ீ peak were significantly lower than those of the pandemic peak no matter in which comparisons (Fig. 3C) . Our results suggested that it took time for finally accumulated Spike mutations to truly and eventually cause mass infection and death in population, like the reported one-month selection time for D614G (3) and matched the selection process in the evolutionary spread (16, 17) . Such time was about six months in our bioinformatics approach. 

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. To estimate selection courses for all continents, we adopted

by pooling all clades including NA data. Similarly, we also observed that COVID-19 data peaks occurred after that of RERS in each continent. This suggested that the selection course we observed using global data also existed in all continents (Fig. 4, Fig. S2 ). Based on the high consistency between the collected SARS-CoV-2 sequence number and the new case number every day (Fig. S3) , there is no system error caused by the data we used here. This on one hand showed a good match with the reported one-month (3) and our model-calculated six-month selection time (Fig. 3A) , and on the other hand clearly illustrated that the length of selection could largely determine when a next pandemic peak could come. Thus, our results suggested that although

could not predict pandemic in a real-time manner, its most recent peaks could serve as an important indicator of the following waves of infections and deaths.

Based on the above results and analysis, we then predicted future pandemic trends based on current COVID-19 data with or without our RERS method (Fig. 5) . Except in Oceania and South America, RERS trends were different from pure regression predictions in all the remaining continents, no matter using data of deaths ( Fig. 5A) or cases (Fig. 5B) . In Africa, Asia, Europe, and North America, both new cases and/or new deaths will continuously climb to a record-high point around July 2021.

During the development of our bioinformatics approach, we observed that clade GR had both the highest genomic diversity percentage by distribution analysis (Fig. 3A) and fastest revolution by 3B) . Although it was the only exception in our finding that contributed most mutations yet wasn't a small portion of SARS-CoV-2, it is still the truth that the latest clade All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint GRY evolved out of it, found in the United Kingdom in December 2020 for the first time (6, 12, 14) and became the largest clade since February 2021. Now that we predicted a future wave using

, we would continue to explore whether these two clades could contribute to new variants which might potentially cause that wave. To do this, we separately analyzed MNPK and 

higher than 0.3 implied relatively high revolution ( Fig. 6B and 6C) . These highly revolutionary variants came from six countries of three continents, and included currently-found variants: P.1 (18, 19) (Fig. 6D and 6E) . These variants of high MNPK and

values positively showed a huge potential of clade GR to develop new variants in the future through continued fast revolution.

On the contrary, B.1.1.7 in clade GRY showed a decelerated evolutionary trend in almost all continents (Fig. 6F) , although all sampled strains from different countries had mutation rates higher than our threshold 14.5 (Fig. 6G) . Together with the result that their

values were all below zero (Fig. 6H) , new variants could not possibly develop out of it.

Conclusively, these results together suggested that sustained Spike revolution could be the most possible determinant of new variant's emerging. This could be explained by the established selection theory that if mutations would benefit the survival of spreading of SARS-CoV-2, then they must be tested through a selection stage in a large population.

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Tracking viral evolutionary pace is critical to better understand how emerged variants could impact global spreading. One famous example is that to understand how the HIV evolves from a next-generation sequencing angle is central to the development of appropriate drugs and vaccines (22, 23) . We thus developed a bioinformatics approach to estimate the revolutionary rate over time by our defined metric, RERS. Distinct from the mutation rate evaluated by nucleotide substitutions/site/year for HIV, RERS focuses on changes or trends in the evolutionary process.

Using RERS, we tracked Spike evolution by every 7 sampling days. Moreover, we developed an evolutionary-spreading model to integrate genomic Spike evolution data, COVID-19 data, related geographical indicators and related metadata. This model predicts a next pandemic peak by uniquely estimating the combined mutation and selection process.

Illustrated idea of our model is based on the combined mutation and selection theory in evolutionary biology (Fig. 7) . We found that the new variants carrying advantageous mutations could not be instantaneously dominant but has to experience a selection in a large population for 1 to 6 months. This course varies in the continents, indicating its relations with other factors, such as isolation policy, population size, and control policy (masking, distancing, and travel constriction). This may be also related to the transmissibility of the emerging variants (6) .

Distinct from existing studies only using the missense mutations at the protein level, we considered all types of mutations for the Spike genomic variations. The goal of our analysis is to estimate to what extent the virus mutates its genome, and to define an accurate mutation rate in terms of genomic single nucleotide variations and indels instead of only examining changes to amino acids. The necessity for including all mutations is also supported by recent discoveries for All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint the genomic mutations in the C terminal of S protein in lineages B.1.1.7 and P.1, which are of unknown significance in associating with S protein functions (12) . Our results confirmed that the virus has a low mutation rate normally but high dynamics when the virus reaches its limit because of the selective pressure. According to MNPK and RERS analysis, B.1.1.7 is the result of dramatically increased evolutionary rate (about 3-fold increased from October 2020 to March 2021) and taking about 6 months from its emergence to the current prevalent spread by a high selection. The emerged variants could be sponsored by an accelerated evolution during which some viruses, not necessarily the largest clade but could be relatively smaller ones, acquired the highest mutation rate.

We also analyzed MNPKs in different gender and age groups, found no difference among genders (Fig. S4) . However, the largest mutations contributed by clades GH, GR, and GRY existed in people of 20-40 and 40-60 age groups compared to others (Fig. S5) . This could explain why B.1.1.7 surges have shifted to younger population (24) .

Genomic surveillance plays a key role in monitoring new variants and understanding the evolutionary spread (25, 26) . Our model has revealed that the mutation and selection are critical to Spike evolution. Through gaining new mutations and surviving from the high selection pressure, current variants had the ability to spread more quickly in population (3, 6, 21) , evade detection by specific diagnostic tests (27) , and potentially evade natural or vaccine-induced immunity (8, 12, 28) . Thus, the model could answer the question why cases and deaths were still increasing while people have been receiving vaccines: SARS-CoV-2 has been experiencing a second evolutionary acceleration since October 2020 and current variants such as B.1.1.7 already succeed selection thus would possibly reach a peak around July 2021. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint So far, we have been equipped with several approved vaccines or candidates under phase 3 (24) .

Though it is urgent to evaluate if they are effective enough to compete Spike evolution, our model cannot make such a prediction due to insufficient vaccine data by March 23 rd , 2021 ( Fig.   S6 ) but this will be our future focus after integrating evolutionary rate, pandemic data, and vaccine data. Future predictions would not only be useful to predict the pandemic trend post vaccination but also to evaluate vaccine effects in the context of Spike evolution.

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

We downloaded the 817,351genome sequences from 6 continents and 185 countries/regions from GISAID (https://www.gisaid.org) as of March 15 th , 2021. We also downloaded the sample metadata from GISAID, including information of geographic location, clade, lineage, collection date, as well as age and gender of confirmed cases. To associate the evolutionary data with the pandemic data, we downloaded the COVID-19 data from the World Health Organization 

We aligned the genomic sequence to the Spike reference from WIV04-reference (13), using the BLAST tools (5). We constructed a genome database for the 817,351genome sequences of SARS-CoV-2, downloaded from GISAID. We used BLAST for nuclei (blastn) as the alignment tool to align the 817,351genome sequences to the WIV04-reference. The newest version of BLASTN was downloaded from the National Institutes of Health BLAST website (https://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/), which was ncbi-blast-2.11.0+.

The parameters used in the alignment are as follows and other parameters were used as default, -perc_identity 90 -max_target_seqs 10000000 -outfmt "6 qseqid sseqid sallacc stitle pident ppos All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint gaps gapopen mismatch length qstart qend sstart send evalue bitscore score" We implemented the large-scale high-performance computing using Google Cloud Computing.

We defined the mutation rate by normalizing the number of mismatches and indels to the length of the Spike genomic region and multiplying 1,000. We denote the number of mismatches and indels as n , and the length of the Spike genomic region as N, and the mutation rate at the genomic level is

where MNPK represents Mutated Number Per Kilobases, evaluating the mutation rate of the Spike genomic region.

To evaluate the mutation rate changes of the virus over time, we defined RERS to quantify the evolutionary rate. The definition of RERS is based on the MNPKs of the virus genomes over time. RERS is quantified by three steps: (1) using a window, e.g., 7 days, to group the genomes;

(2) calculating average MNPK for each group of genomes; and (3) (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. To show if the evolutionary rate is increased or decreased, we deducted RERS by 1.

Thus, a positive value of RERS indicates a speed-up evolution whereas a negative RERS suggests a decreased pace of evolution.

To identify the trends and waves in the evolutionary rates and COVID-19 data, we applied the LOWESS regression (29) to the RERS data and the Gamma regression (30) 

The loss function is described as:

Gamma regression is parameterized in terms of a shape parameter

and an inverse scale

, thus a Gamma distribution can be described as All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint

The probability density function in the shape-rate parameterization is

To correlate the changes in the evolutionary rates and the corresponding changes in the pandemic data, we removed the course between the peak of RERS and the peak of the COVID-19 data and did data reshape based on the widths of the peaks in the RERS and the COVID-19 data (Fig. S2) .

We considered the COVID-19 pandemic data of new_deaths_per_million and new_cases_per_million. The relation between the evolutionary data (RERS) and the COVID-19

data is based on the mutation, selection, and spread processes associated with new variants, in which the incurred new cases and new deaths happen after the processes of mutation and selection of the new variants. We selected the latest peak and/or valley from a RERS distribution of a continent for association with the COVID-19 data. The RERS data from 200-450 days after Jan 1 st , 2020 in Africa, Europe, South America, and Oceania, and the RERS data collected 100-450 days after Jan 1 st , 2020 from Asia and North America were considered in the correlation analyses of the COVID-19 data.

After removing the course between the peak of RERS and the peak of the COVID-19 data, the peak of RERS can show consistent trends with that of the COVID-19 data. To find out the underlying association between these two types of data, we rescaled the peak of RERS by its All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint width and that of the COVID-19 data. To accurately identify the widths of the peaks, we selected the summit point of the peak, ܲ , and the lowest point of the adjacent valley, ܸ (Fig. S2) . Thus, the width of a peak ݆ is defined by

We reshaped the RERS peak by the peak width of RERS,

and that of the COVID-19 data,

. The transformed data of RERS distribution were defined by the course between the RERS peak and the COVID-19 data peak,

The correlations between the RERS and COVID-19 data were enabled by the reshaped distribution of RERS considering the peak widths. The statistical significances of the correlations were calculated by the reshaped distributions ‫ݕ(‬ ᇲ ) and the distribution of the COVID-19 data on the same days.

We predicted the future trends of new cases and new deaths till July 3 rd , 2021. Each prediction provides two trends: one boundary regressed by only COVID-19 data and another is the regression from the combined RERS and COVID-19 data.

The identified highly statistically significant correlation between RERS and the COVID-19 data provided the basis for our predictive model. The significant correlation suggests that the days All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. ; https://doi.org/10.1101/2021.05.06.21256705 doi: medRxiv preprint used for mutation and selection may determine the spans of increase and decrease of the upcoming wave in the COVID-19 data that follows the mutation and selection processes.

For the prediction purpose, we not only reshaped the RERS data by removing the course between the peaks,

and rescaling the RERS data by the peak widths, but also reshape the RERS data by the summit values of the two peaks. We denoted the summit value of the RERS

and that of the COVID-19 data as

. We rescaled the ‫ݕ‬ ᇲ as follows.

We used the Gamma regression to predict the future trends of the pandemic. Since our prediction is based on the course between the peaks of RERS and COVID-19 data ‫ܥ(‬ 

The authors declare that they have no competing financial interests.

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 May 10, 2021. 

COVID-19 CG enables SARS-CoV-2 mutation and lineage tracking by locations and dates of interest

Potentially adaptive SARS-CoV-2 mutations discovered with novel spatiotemporal and explainable AI models

Tracking Changes in SARS-CoV-2 Spike: Evidence that D614G Increases Infectivity of the COVID-19 Virus

Emergence of SARS-CoV-2 through Recombination and Strong Purifying Selection. bioRxiv

Origin-independent analysis links SARS-CoV-2 local genomes with COVID-19 incidence and mortality

Estimated transmissibility and impact of SARS-CoV-2 lineage B.1.1.7 in England

Structure-Function Analyses of New SARS-CoV-2 Variants B.1.1.7, B.1.351 and B.1.1.28.1: Clinical, Diagnostic, Therapeutic and Public Health Implications

B.1.526 SARS-CoV-2 variants identified in New York City are neutralized by vaccine-elicited and therapeutic monoclonal antibodies. bioRxiv

Evidence of escape of SARS-CoV-2 variant B.1.351 from natural and vaccine-induced sera

Spike mutation D614G alters SARS-CoV-2 fitness

Covid-19: The E484K mutation and the risks it poses

Emerging SARS-CoV-2 Variants and Impact in Global Vaccination Programs against SARS-CoV-2/COVID-19. Vaccines (Basel) 9

Different mutations in SARS-CoV-2 associate with severe and mild outcome

Sensitivity of SARS-CoV-2 B.1.1.7 to mRNA vaccine-elicited antibodies

Emergence of SARS-CoV-2 through recombination and strong purifying selection

Natural selection in the evolution of SARS-CoV-2 in bats created a generalist virus and highly capable human pathogen

On the evolutionary epidemiology of SARS-CoV-2

Imported SARS-CoV-2 Variant P.1 in Traveler Returning from Brazil to Italy

Rapid Detection of SARS-CoV-2 Variants of Concern, Including B.1.1.28/P.1, in British Columbia

Introduction of Brazilian SARS-CoV-2 484K.V2 related variants into the UK

Genomics and epidemiology of a novel SARS-CoV-2 lineage in Manaus

Recent advances in understanding HIV evolution. F1000Res

The evolution of HIV-1 and the origin of AIDS

The first 12 months of COVID-19: a timeline of immunological insights

Tracking the emergence of new SARS-CoV-2 variants in South Africa

Genomic surveillance reveals multiple introductions of SARS-CoV-2 into Northern California

Could mutations of SARS-CoV-2 suppress diagnostic detection?

Infection-and vaccine-induced antibody binding and neutralization of the B.1.351 SARS-CoV-2 variant

Graphical assessment of internal and external calibration of logistic regression models by using LOWESS smoothers

Comparison of statistical and machine learning models for healthcare cost data: a simulation study motivated by Oncology Care Model (OCM) data. BMC

The authors of this study would like to acknowledge all the researchers who openly shared their genomic data on GenBank and GISAID. We also acknowledge the support of the Wake Forest

All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.