In December 2020, routine genomic surveillance in the United Kingdom (UK) ¹ reported a new and genetically distinct phylogenetic cluster of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (variant VOC202012/01, lineage B.1.1.7). Preliminary analysis suggests that this lineage carries an unusually large number of genetic changes ² . The earliest known cases of B.1.1.7 were sampled in southern England in late September 2020, and by December the lineage had spread to most UK regions and was growing rapidly ³ . In October 2020, a separate SARS-CoV-2 cluster (variant 501Y.V2, lineage B.1.351), which carried a different constellation of genetic changes, was detected by the Network for Genomic Surveillance in South Africa ^{4, 5} . Both lineages carry mutations, especially in the virus spike protein, that may affect virus function, and both appear to have grown rapidly in relative frequency since their discovery. Early analyses of the spatial spread of SARS-CoV-2 highlights the potential for rapid virus dissemination through national and international travel ^{6, 7} . Therefore continued genomic monitoring of lineages of concern is required.

To facilitate tracking of these lineages on an international scale, we developed a software tool grinch (global report investigating novel coronavirus haplotypes) that collates SARS-CoV-2 genomic data and epidemiological metadata. Resources such as grinch on cov-lineages.org can inform public health bodies and institutions around the world. Other excellent resources to track lineages and variants are available, including covariants.org, which tracks the spread of SARS-CoV-2 variants of interest, and outbreak.info, which gathers multiple sources of genetic and epidemiological data to track lineages. We include a non-exhaustive list of resources for tracking SARS-CoV-2 at https://cov-lineages.org/resources.html.

To better characterise the international distribution of lineages B.1.1.7 and B.1.351 we sourced SARS-CoV-2 sequences from GISAID ^{8, 9} and assigned lineages using pangolin (v2.1.6, https://github.com/cov-lineages/pangolin), which implements the nomenclature scheme described in Rambaut et al., ¹⁰ . Genomes are assigned lineage B.1.1.7 if they exhibit at least 5 of the 17 mutations inferred to have arisen on the phylogenetic branch immediately ancestral to the cluster ( Table 1) ² ; or to B.1.351 if they exhibit at least 5 of 9 lineage-associated mutations ( Table 1) ⁵ . Lineage count and frequency data have been calculated daily using grinch. Using International Air Transport Association (IATA) travel data from October 2020, available through bluedot.global, we aggregated and collated the passenger volumes from international airports in London and South Africa to international destinations on same booking. Destinations with more than 5,000 passengers from London and more than 300 passengers from South Africa during the month of October are displayed on the cov-lineages.org website and in the underlying data for this publication ¹¹ . grinch, with custom python modules that make use of geopandas v0.9, matplotlib v3.2 and seaborn v0.10, combines this information and produces reports with descriptive tables and figures that can be found at https://cov-lineages.org/global_report.html.

As of 7th Jan 2021, 45 countries had reported the presence of B.1.1.7 and 13 countries had reported B.1.351/501Y.V2. B.1.1.7 and B.1.351 genome sequences were available for 28 and 8 countries, respectively ( Figure 1a, b, c) ¹¹ . Although some countries report increases in the relative frequency of B.1.1.7, genome sequencing efforts vary considerably. Potential targeting of sequencing towards travelers from the UK could bias frequency estimates upwards ( Figure 1b, c) and differing genome sharing policies and delays may also skew reporting estimates. The time between the initial collection date of a new variant sample in a country and the first availability of a corresponding virus genome on GISAID was, on average, 12 days (range 1–71).

The number of B.1.1.7 and B.1.351/501Y.V2 genome sequences reported in each country is a consequence of (i) the intensity of local genomic surveillance; (ii) the level of concern about new variant introductions; (iii) the volume of international travel among affected countries, and (iv) the amount of local transmission following the introduction of lineage from elsewhere. To explore these factors, we analysed the most recent available IATA travel data (October 2020). We collated the total number of origin-to-destination air journeys between major London international airports and each country. The calculation was repeated for journeys originating in all international South African airports. We focussed on London and South Africa as they are the locations with the first reports and highest reported prevalence of lineages B.1.1.7 and B.1.351 respectively ^{2, 5} . However, due to low SARS-CoV-2 genomic surveillance in many locations, we cannot reject the hypotheses that these lineages initially originated elsewhere. Figure 1d shows destinations receiving >5,000 travellers in October 2020 from the UK ( Figure 2 shows destinations receiving >300 travellers from South Africa).

Of the countries that receive >5,000 travellers from London, 16 have sequenced B.1.1.7. Of the 45 countries that have identified B.1.1.7 (32 in travellers and 13 with local onward transmission), only 6 perform real-time routine genomic surveillance (Denmark, UK, Iceland, The Netherlands, Australia, Sweden), 3 have prioritised sequencing based on S-gene target failure tests ¹⁶ , 30 primarily targeted sequencing towards arriving travellers from the UK, and there was no information available for 10 (details at https://github.com/cov-lineages/lineages-website/blob/master/_data/). Of the 13 countries that have identified B.1.351 (four with local onward transmission including South Africa), 4 perform routine sequencing (South Africa, UK, Botswana, Australia), 6 target sequencing of travellers, and there was no information available for 3. Consequently, there is no clear relationship between number of sequences reported and flight numbers, but rather reflects the current genomic surveillance effort. For example, in September, the UK sequenced ~13% of its reported cases and Denmark sequenced ~21%. In comparison, Israel sequenced ~0.002% of its cases during the same period ^{17, 18} .

Our study has several limitations. The passenger flight data do not include recent changes to holiday travel, and recent restrictions on travel from the UK and South Africa is not reflected in the mobility data. Further, flight data may not accurately reflect the final destination if multiple tickets are purchased.

The discovery and rapid spread of B.1.1.7 and B.1.351/501Y.V2 highlights the importance of real-time and open data for tracking the spread of SARS-CoV-2 and for informing future public health interventions and travel advice.