key: cord-0802566-hde742ev
authors: Brunner, F. S.; Brown, M. R.; Bassano, I.; Denise, H.; Khalifa, M. S.; Wade, M.; Kevill, J. L.; Jones, D. L.; Farkas, K.; Jeffries, A. R.; The COVID-19 Genomics UK Consortium,; Cairns, E.; Wierzbicki, C.; Paterson, S.
title: City-wide wastewater genomic surveillance through the successive emergence of SARS-CoV-2 Alpha and Delta variants
date: 2022-02-16
journal: nan
DOI: 10.1101/2022.02.16.22269810
sha: 49d994a6b53fbab6d96340d3eeb6dc3a6497aaa6
doc_id: 802566
cord_uid: hde742ev

Genomic surveillance of SARS-CoV-2 has been essential to provide an evidence base for public health decisions throughout the SARS-CoV-2 pandemic. Sequencing data from clinical cases has provided data crucial to understanding disease transmission and the detection, surveillance, and containment of outbreaks of novel variants, which continue to pose fresh challenges. However, genomic wastewater surveillance can provide important complementary information by providing estimates of variant frequencies which do not suffer from sampling bias, and capturing all variants circulating in a population. Here we show that genomic SARS-CoV-2 wastewater surveillance can detect fine-scale differences within urban centres, specifically within the city of Liverpool, UK, during the emergence of Alpha and Delta variants between November 2020 and June 2021. Overall, the correspondence between wastewater and clinical variant frequencies demonstrates the reliability of wastewater surveillance. Yet, discrepancies between the two approaches in when the Alpha variant was first detected emphasises that wastewater monitoring can also capture missing information resulting from asymptomatic cases or communities less engaged with testing programmes, as found by a simultaneous surge testing effort across the city.

Genomic surveillance has been a significant feature in the public health response to the SARS-CoV-2 pandemic (Wu et al. 2020; Zhu et al. 2020 ) because of its ability to 45 detect the emergence of and track new variants of concern (VOC) (Robishaw et al. 2021; Vöhringer et al. 2021) . Important examples include the B.1.1.7 (Rambaut et al. 2020 ), B.1.351 (Tegally et al. 2020) , P.1 (Faria et al. 2021 ), B.1617.2 (Cherian et al. 2021 ) and, most recently, the B.1.1.529 lineage (Qin et al. 2021) , named VOC Alpha, Beta, Gamma, Delta, and Omicron, respectively. These VOC have demonstrated one 50 or more attributes out of increased transmissibility, more severe disease, a reduction in antibody neutralisation, reduced therapeutic response or reduced vaccine effectiveness (Davies et al. 2021; Robishaw et al. 2021) . Thus, while vaccination currently provides substantial protection against all known VOC, continued genomic surveillance is essential to mitigate and contain the threat they pose to public health. 55

It informs the implementation and assessment of non-pharmaceutical interventions (e.g. social distancing, lockdowns and regional, national and international restrictions) and targeted surge testing, and serves as an early warning system for the emergence and spread of novel variants Tegally et al. 2020) .

Genomic surveillance of SARS-CoV-2 has primarily been driven by whole genome 60 sequencing of clinical isolates, typically using residual RNA from diagnostic RT-qPCR tests. One million SARS-CoV-2 genomes were sequenced worldwide by April 2021, rising to over seven million by January 2022 on the GISAID database (Elbe and Buckland-Merrett 2017) . This has provided unprecedented insight into the joint evolution and epidemiology of the SARS-CoV-2 pandemic (Ward et al. 2021; Harvey 65 et al. 2021) . Nevertheless, the cost of clinical sequencing to generate these data has been, and continues to be, substantial (10 -35 GBP per sample for consumables (Tyson et al. 2020) , plus similar costs for staff, logistics and data infrastructure). It may be unsustainable at the levels required to adequately inform public health authorities as SARS-CoV-2 becomes endemic and threatens public health for the 70 foreseeable future, even in developed nations.

Wastewater-based surveillance is a complementary, cost-effective approach to clinical sequencing, which has gained significant attention throughout the COVID-19 pandemic (Jahn et al. 2021; Hillary et al. 2021; Peccia et al. 2020; Smyth et al. 2021; Mishra et al. 2021) . Given that SARS-CoV-2 is shed in faeces by more than 50% of 75 infected people (Foladori et al. 2020) , it can be recovered from wastewater, its RNA extracted, and its presence and quantity in a wastewater catchment determined using RT-qPCR (Farkas et al. 2020) , with trends generally tracking the rise and fall of corresponding clinical cases (Peccia et al. 2020; Hillary et al. 2021; Wade et al. 2020 ). This can be achieved for entire populations by sampling at the inlet of 80 wastewater treatment plants, or at much finer spatial scales, such as across cities, by sampling within the sewer network.

More recently, the recovery of SARS-CoV-2 genomes from wastewater has opened up the possibility of detecting and tracking circulating SARS-CoV-2 variants (Jahn et al. 2021; Hillary et al. 2021; Peccia et al. 2020; Smyth et al. 2021; Brown et al. 2021) . 85

Such an approach is particularly attractive for population-level insights during periods of high prevalence, especially if capacity constraints reduce the proportion of sequenced positive RT-qPCR tests. Furthermore, it can detect asymptomatic cases and is proposed to capture communities under-represented by clinical testing, particularly in urban centres (Polo et al. 2020; Green et al. 2021) . 90

Nevertheless, moving from detection and quantification of SARS-CoV-2 in wastewater by RT-qPCR to characterisation by genome sequencing is challenging.

. CC-BY-NC 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 February 16, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 The low abundance of SARS-CoV-2 means enrichment through RNA concentration methods is necessary, simultaneously enriching PCR inhibitors and contaminating bacterial, viral, and human nucleic acids (Peccia et al. 2020) . SARS-CoV-2 genomes 95 in wastewater are also highly degraded and fragmented. In combination, this can result in poor and inconsistent amplification of target amplicons and thus patchy genome coverage. Even if amplification and sequencing are successful, data interpretation can be difficult. Wastewater harbours a mixed SARS-CoV-2 population. Therefore, sequences are derived from a pool of fragments, removing much of the 100 phase information between polymorphic sites on the genome used to assign phylogeny and subsequently lineage. However, by reference against clinicallyderived genomes of known SARS-CoV-2 lineages, wastewater data has the potential to detect and quantify polymorphisms characteristic of defined lineages and of VOC in particular (Jahn et al. 2021; Fontenele et al. 2021) . 105

This study demonstrates the utility of wastewater-based genomic surveillance of SARS-CoV-2 using longitudinal data collected from multiple locations in a single city -Liverpool, UK -between November 2020 and June 2021. During this time, Liverpool was the subject of a pilot study evaluating lateral flow tests for rapid asymptomatic testing . This pilot noted the link between 110 social inequalities and testing uptake, with social deprivation and digital exclusion as major factors limiting uptake (Green et al. 2021) . Wastewater-based epidemiology can provide valuable insight into some of the communities or areas of Liverpool that may be less accessible to conventional testing. This period in the UK also saw the emergence and establishment of Alpha (B.1.1.7) and subsequently Delta (B.1.617.2) 115 SARS-CoV-2 variants (Vöhringer et al. 2021) . We show that wastewater genomic surveillance can reliably detect the emergence of these variants and their subsequent rise across a city.

. CC-BY-NC 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) (Fig. 1, Table S1 ). In addition, concurrent samples from four WWTPs in the southeast of England were collected between 2 nd of September 2020 and 17 th of January 2021 as a control group. Samples were transported and subsequently stored Extracts were stored at -80°C until further processing.

Wastewater RNA extracts were purified and sequenced with a standardised EasySeq™ RC-PCR SARS CoV-2 (Nimagen) V1.0 protocol (Jeffries et al. 2021 ). In 145 short, samples were cleaned with Mag-Bind® TotalPure NGS beads (Omega Bio-Tek) and then reverse transcribed using LunaScript® RT SuperMix Kit (New England . CC-BY-NC 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 February 16, 2022. ; https://doi.org/10. 1101 /2022 Biolabs) and the EasySeq™ RC-PCR SARS CoV-2 (novel coronavirus) Whole Genome Sequencing kit v3.0 (NimaGen). Amplicons were pooled and libraries cleaned with Mag-Bind® Total Pure NGS beads (Omega Bio-Tek) before sequencing 150 on an Illumina NovaSeq™ 6000 platform generating 2x150bp paired end reads.

We processed raw reads using the ncov2019-artic-nf v3 pipeline (https://github.com/connor-lab/ncov2019-artic-nf) using default parameters. Briefly, reads were aligned to the SARS-CoV-2 reference genome (MN908947.3, (Wu et al. 155 2020)) using the Burrow-Wheeler Aligner (bwa) (Li and Durbin 2009) . We then identified Single Nucleotide Polymorphisms (SNPs) and insertions/deletions (Indels) from BAM files using samtools (v1.13, (Danecek et al. 2021) ) and VarScan (v2.4.4, P < 0.05, all other settings default, (Koboldt et al. 2012 )) on 100,000 sequencing reads with an alignment score > 10. Next, we filtered the identified SNPs and Indels against 160 signature mutations of known VOC and variants under investigation (VUI), as defined by Public Health England (PHE) at the time of writing (https://github.com/phegenomics/variant_definitions, Table S2 , Fig. 2 ). We further used custom scripts to extract summary statistics and sequence quality indicators, such as genome coverage, mapped reads and read depth (Table S3) . 165

To aid VOC and VUI identification at low frequencies from wastewater samples, we adopted a recently described amplicon-level co-occurrence approach (Jahn et al. 2021 ). Briefly, co-occurring mutations were called from BAM files using CoOccurrence adJusted Analysis and Calling (COJAC) (Jahn et al. 2021) , facilitating the identification of signature mutations co-occurring on the same sequencing read, 170 that is, a read or paired read coming from the same amplicon, thus one SARS-CoV-2 virion. This greatly improves confidence in variant detection, especially at low frequencies, since co-occurring mutations are less likely to arise through sequencing error than individual SNPs (Jahn et al. 2021) . We extracted co-occurrence signature . CC-BY-NC 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 for final model structures), as indicated by Akaike Information Criterion (AIC) and likelihood ratio tests (Cribari-Neto and Zeileis 2010). To account for missing data in SNP/indel frequencies, a weighting factor was applied using the number of used signature SNPs/indels (weights = n). We assessed model validity by 200 visual checks of homoscedasticity of the standardised weighted residuals and . CC-BY-NC 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 February 16, 2022. ; https://doi.org/10.1101/2022.02.16.22269810 doi: medRxiv preprint linearity of the model fit (Fig. S2) . We then extracted likelihood ratio tests of estimated marginal means for each predictor variable (joint_tests, "emmeans" (Lenth 2021) , Table S4 ).

We also compared the frequency of detected VOC/VUI signature SNPs/indels in 205 wastewater samples to the frequency of VOC/VUI identified in clinical cases by the 

Across all catchments, we observed a significant increase in the mean frequency of Alpha (B.1.1.7) signature SNPs/indels between the 2 nd November 2020 and the 28 th 220

February 2021 (F 1 = 13667, P < 0.001, Fig. 3 , Table S4 ). This closely corresponds with the observed rise in Alpha clinical cases across Liverpool for the same period ( Fig. 4) and wastewater data from four WWTPs in the southeast of England (F 1 = 13829, P < 0.001, Table S4 , Fig. S3 ). For most sites, the rise of Alpha began in mid to late December, with peak frequencies observed in late January and early February 225 ( Figs. 3 and 4) . As defined by co-occurrence analysis, the earliest wastewater Alpha detection preceded clinical detections in five of the nine sites by up to 55 days (Fig.   4) . The contrary was observed in the remaining four sites, with clinical samples . CC-BY-NC 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 February 16, 2022. ; https://doi.org/10.1101/2022.02.16.22269810 doi: medRxiv preprint picking up Alpha up to 26 days earlier (Fig. 4) .

We detected local differences between wastewater catchments (date: site interaction, 230 F 8 = 32.8, P < 0.001, Table S4 ). This was most notable in Strand SSO (STS), where we observed a high frequency of Alpha signature SNPs from four samples in early November (Fig. 3) , though this signal diminished before further detections in early January. Similarly, we observed Alpha signature SNPs at a low to moderate frequency at Fazakerley High (FZH) as early as mid-November, while they were 235 barely detected until late December in Mersey Road (MRD, Fig. 3 ). This suggests Alpha spread through parts of the north of the city earlier than through the south (Fig.   1 ), a finding corroborated by co-occurrence analysis but not clinical data (Fig. 4) . 3 and 4) . It is noteworthy that the observed transition from Alpha to Delta in wastewater was abrupt (Figs. 4, 5 and S4) . From April to early June, infection 245 numbers were low across Liverpool (Fig. S5) , and wastewater SARS-CoV-2 concentrations were consequently low (J. Kevill & D. Jones, unpublished data) . This is reflected in the observed reduction in mapped reads and genome coverage for this period (Fig. S1) . Indeed, the detection of Alpha and Delta signature SNPs was more sporadic during this period (Figs. 4, 5 and S4 ). Co-occurrence analysis was not 250 informative for the Delta variant since the only pair of signature mutations cooccurring on the same amplicon of our panel is not unique to the Delta variant but also part of the B.1.629 and AY.3 variants, among others. This explains why we detected a co-occurrence signal as early as November for all sites (Fig. S4) .

Finally, we also observed signature SNPs of other VOC and VUI in wastewater 255 . CC-BY-NC 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 February 16, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 samples across Liverpool (Fig. S6) , however detections were sporadic and at low predicted frequencies, which corresponds to clinical cases. Although there were notable exceptions, including the B.1.351 (Beta), P.1 (Gamma), P.3 (Theta) and B.1.1.318 lineages (Fig. S6 ).

Genomic surveillance of wastewater has already shown great promise throughout the unfolding SARS-CoV-2 pandemic (Polo et al. 2020; Farkas et al. 2020) , including the detection of VOC (Fontenele et al. 2021) , in some cases prior to clinical detection (Jahn et al. 2021) . Here, we have demonstrated that wastewater monitoring can also reveal fine-scale, local differences in the spread of VOC across urban centres. As seen for the Alpha variant here and by Jahn et al. (2021) , genomic surveillance of wastewater can detect VOC earlier than clinical testing. In both instances, cooccurrence analysis improved confidence in (early) low-frequency variant detection by identifying multiple linked mutations from the same virion instead of solely relying on single signature mutations. This requires the co-occurrence of mutations unique to 275 a given variant on amplicons of the used sequencing scheme. When no unique mutation set is available, as is the case for the Delta variant and the NimaGen SARS-COV-2 whole-genome sequencing kit used here, reliable variant detection via co-occurrence analysis is not possible. The software developers have acknowledged this limitation, and the design of primers to create appropriate co-occurrence 280 amplicons for relevant sequencing schemes is suggested as a workaround (Jahn et al. 2021) . It is important to note that even in cases where co-occurrence analysis is applicable, our fine-scale local data highlighted that wastewater monitoring . CC-BY-NC 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 February 16, 2022. ; https://doi.org/10.1101/2022.02.16.22269810 doi: medRxiv preprint sometimes detects new variants earlier than clinical testing, but not always. The reasons for this are yet unclear. It is likely that the inherent variability of wastewater 285 detections, due to variations in viral shedding rates and dilution from rainfall, etc (Polo et al. 2020 Indeed, our finding that Alpha was detected in wastewater in North Liverpool much earlier than clinical cases had indicated, corresponds well with findings from a largescale asymptomatic testing campaign, which found that testing uptake was lower in North Liverpool, yet the rate of positive tests higher (Green et al. 2021) . Clearly, a 300 combination of genomic surveillance of clinical cases and wastewater is most likely to detect new variants as early as possible and provides the most precise picture of unfolding variant dynamics to inform public health measures ).

Intriguingly, when comparing peak Alpha and Delta frequencies in corresponding clinical and wastewater data, we find that each variant, in turn, reaches complete 305 dominance, with estimated frequencies of 100% and ~75% in clinical and wastewater samples, respectively. This suggests improved statistical methods may be required to estimate lineage frequencies from pooled sequencing data, as produced from wastewater samples. Here we have relied on a relatively crude estimation by taking the mean of signature SNP/Indel frequencies. However, genetic variation may mean 310 that a SNP/Indel may not be present on all branches within a lineage, whereas a . CC-BY-NC 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 February 16, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 single fully phased viral genome from a clinical sample would be reliably assigned to a lineage. Following methods for analysing metagenomic amplicon data (Quince et al. 2011) , the development of statistical methods to infer lineage proportions from multiple amplicons, while controlling for sequencing error may be productive. 315

The limited quality of sequences obtained from wastewater during April and May 2020 also highlights the current limits of variant detection via wastewater sequencing when case numbers, and hence SARS-CoV-2 concentrations, are low. While the rise of Delta was evident in our results (Fig. 5) , the transition from Alpha to Delta, compared to the gradual emergence of Alpha, was less visible and more abrupt. The 320 use of a relatively short amplicon scheme, as used here, may help to boost sensitivity given the likely degradation of SARS-CoV-2 RNA in wastewater; and from initial work gives better genome coverage than the longer ARTIC v3 amplicons more widely used for sequencing SARS-CoV-2 (data not shown). As previously noted, however, the use of short amplicons limits the use of co-occurrence analysis for variant 325 detection. It is also anticipated that the increased adoption of wastewater-based epidemiology will drive innovation in wastewater sampling, concentration and RNA extraction, improving viral qPCR and sequencing sensitivity (Polo et al. 2020; Hillary et al. 2021; Kevill et al. 2022 ).

In conclusion, we show that wastewater genomic sequencing can detect the 330 emergence and rise of new SARS-CoV-2 variants. Results correspond well with those obtained through genomic sequencing of clinical samples. In some cases, variants are observed prior to clinical detections, which may be particularly useful in areas or communities with low testing uptake. 

Wastewater sequencing data will be deposited and publicly available on the European Nucleotide Archive by 1 st July 2022. The clinical case data used in this study is visualised at https://www.cogconsortium.uk/tools-analysis/public-dataanalysis-2/. A filtered, privacy conserving version of the lineage-LTLA-week dataset is publicly available online (https://covid19.sanger.ac.uk/downloads) and gives 350 access to almost all used data, despite a small number of cells having been suppressed to conserve patient privacy.

Use of surplus nucleic acid derived from routine diagnostics and associated patient data was approved through the COG-UK consortium by the Public Health England 355

Research Ethics and Governance Group (R&D NR0195).

. CC-BY-NC 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 February 16, 2022. ; Table S1 for further catchment details. 365 Figure 2 . SARS-CoV-2 genome structure and signature mutations of VOC Alpha and Delta. Black stars show unique mutations for the Alpha or Delta variant. White stars show mutations shared with at least one other VOC or VUI and therefore not used for mean frequency estimates. Details of all mutations can be found in Table S2 . Genome structure adopted from (Wu et al. 2020 ).

. CC-BY-NC 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 February 16, 2022. ; https://doi.org/10.1101/2022.02.16.22269810 doi: medRxiv preprint (Table S4 ).

Point shape indicates the number of unique alpha mutations used in the mean calculation for a given sample: empty circles: 1 mutation, crossed square: 2 to 5 mutations, filled circles: >5 mutations.

. CC-BY-NC 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 February 16, 2022. ; https://doi.org/10.1101/2022.02.16.22269810 doi: medRxiv preprint Figure 4 . Mean frequency of B.1.1.7 (Alpha) signature SNPs/Indels detected in wastewater versus clinical samples in each catchment, 2 nd November 2020 to 26 th June 2021. Points show the mean frequency of unique Alpha mutations for a given wastewater sample (blue) and the frequency of Alpha clinical cases from a given date (yellow). Coloured lines show the respective local polynomial regression fit including shaded 95% confidence intervals. Vertical 385 lines indicate the first confirmed clinical case of Alpha (yellow) and the first wastewater detection of co-occurring Alpha mutations on amplicon 147 (blue). (Table S4) . Point shape indicates the number of unique alpha mutations used in the mean calculation for a given sample: empty circles: 1 mutation, crossed square: 2 to 5 mutations, filled circles: >5 mutations.

Wastewater 400 Monitoring of SARS-CoV-2 Variants in England: Demonstration Case Study for Bristol

Convergent Evolution of SARS-CoV-2 Spike Mutations, L452R, E484Q and P681R, in the Second Wave of 405 COVID-19 in Maharashtra

Beta Regression in R'

Twelve Years of SAMtools and BCFtools

Estimated 415 Transmissibility and Impact of SARS-CoV-2 Lineage B.1.1.7 in England

Data, Disease and Diplomacy: GISAID's Innovative Contribution to

Genomics and Epidemiology of the P.1 SARS-CoV-2 Lineage in Manaus, Brazil'

Wastewater and Public Health: The Potential of Wastewater Surveillance for Monitoring COVID-19'. Current Opinion in Environmental Science &

Concentration and Quantification of SARS-CoV-2 RNA in Wastewater Using Polyethylene Glycol-Based Concentration and QRT-PCR'

SARS-CoV-2 from Faeces to Wastewater Treatment: What Do We Know? A Review

High-Throughput Sequencing of SARS-CoV-2 in Wastewater Provides Insights into Circulating Variants'

Antigen Rapid Lateral Flow Test in the Liverpool Asymptomatic Testing Pilot: Population Based Cohort Study

Evaluating Social and Spatial Inequalities of Large Scale Rapid Lateral Flow SARS-CoV-2 Antigen Testing in COVID-19 Management: An 455 Observational Study of Liverpool

SARS-CoV-2 Variants, Spike Mutations and Immune Escape'

Monitoring SARS-CoV-2 in 465

Municipal Wastewater to Evaluate the Success of Lockdown Measures for Controlling COVID-19 in the UK'

470 'Detection and Surveillance of SARS-CoV-2 Genomic Variants in Wastewater

Wastewater Sequencing Using the EasySeq TM RC-PCR SARS CoV-2 (Nimagen) V1.0'. Protocols

A Comparison of Precipitation and Filtration-Based SARS-CoV-2 Recovery Methods and the Influence of Temperature, Turbidity, and Surfactant Load in Urban Wastewater

VarScan 2: Somatic Mutation and Copy Number Alteration Discovery in Cancer by Exome Sequencing

Emmeans: Estimated Marginal Means, Aka Least-Squares Means (version R package version 1.7.1-1)

Fast and Accurate Short Read Alignment with 490 Burrows-Wheeler Transform

Changing Composition of SARS-CoV-2 Lineages and Rise of Delta Variant in England

CLIMB-COVID: Continuous Integration Supporting Decentralised Sequencing for 500 SARS-CoV-2 Genomic Surveillance'

Measurement of SARS-CoV-2 RNA in Wastewater Tracks Community 505 Infection Dynamics'

Making Waves: Wastewater-Based Epidemiology for COVID-19 -Approaches and 510 Challenges for Surveillance and Prediction'

Genome Characterization and Potential Risk Assessment of the Novel SARS-CoV-2 Variant Omicron

Removing Noise From Pyrosequenced Amplicons'

R: A Language and Environment for Statistical Computing 520 (version 4.1.1). Vienna, Autria: R Foundation for Statistical Computing

Preliminary Genomic Characterisation of an Emergent SARS-CoV-2 Lineage in the UK Defined by 525 a Novel Set of Spike Mutations

Genomic 530 Surveillance to Combat COVID-19: Challenges and Opportunities

RStudio: Integrated Development Environment for R (version 1.4.1717)

Emergence and Rapid Spread of a New Severe Acute Respiratory Syndrome-Related Coronavirus 2 (SARS-CoV-2) Lineage with Multiple Spike Mutations in South Africa

Improvements to the ARTIC Multiplex PCR Method for SARS-CoV-2 Genome Sequencing Using Nanopore

Genomic Reconstruction of the SARS-CoV-2 Epidemic in England'

Wastewater COVID-19 Monitoring in the 555 UK

Growth, Reproduction Numbers and Factors Affecting the Spread of SARS-CoV-2 Novel Variants of Concern in the UK from

Ggplot2: Elegant Graphics for Data Analysis

A New Coronavirus Associated with Human Respiratory Disease

A Novel Coronavirus from Patients with Pneumonia in China

We thank R. Crompton, B. Jones, G. Airey, T. Foster, N. Kadu, C. Nelson and A.Lucaci for help with processing samples and sequencing and United Utilities for providing wastewater catchment mapping data. Funding was provided by NERC