By naming 'a variant' we are flagging it as being of concern even if it probably isn't (relative to other circulating lineages).

Back in April proposed a lineage naming system ( https://doi.org/10.1038/s41564-020-0770-5). We also produced a website, https://cov-lineages.org/  to help track these and they are automatically assigned when viruses are uploaded to #GISAID.

Our system attempts to capture and convey the relationships and history of virus genomes to help with the epidemiology and tracking of the virus and was not intended to say anything about their mutational composition or properties.

Trying to combine these two enterprises is a fool's errand. Here is an example why: There is a Finnish genome that is of lineage B.1.351 (which is mostly 'variant' 501Y-V2) but has P681H (a defining mutation of lineage B.1.1.7 which is mostly variant VOC202012/01).

And there are 4 UK genomes in lineage B.1.1.7 that have E484K (as seen in both B.1.351 and P.1 - the lineage recently described in Brazil). Each of those probably arose independently but are the same variant (and the Finnish one is almost the same variant as those 4 too).

My suggestion for this is not to even attempt to name variants (the term just means genetically distinct). Instead describe constellations or combinations of mutations that we are concerned about and track when and where we see them irrespective of the origin of the lineage.

We should be talking about the frequency of a constellation (e.g., 484K-501Y-681H) for which we have good empirical or theoretical evidence may be of concern or importance. Another virus with the same constellation would be equally concerning irrespective of where it originated.

Particular lineages (such as B.1.1.7) may be rapidly increasing in frequency and it may be due to the constellation of mutations they have - or it may be due to the particular epidemiology of where they are spreading.

Separating out the lineage and clusters (viruses connected by being in the same transmission networks) from mutation constellations (viruses connected by likely having similar properties) will be critical going forward.