A population geneticist made this analogy to me some time ago, and I think it is apt here.
Imagine a drunk person at a street corner. The distance that drunk person moves from the street corner represents the rise of new genetic variants. If there is a slight grade to that street (a genetic advantage), the drunk will tend, over time, to move downhill. But even if there is no grade to the street, over time the drunk will through chance, wind up some distance away from their original position.
The bottom line is that if you had a mutation that made the virus more transmissible, you would see it slowly become more prevalent. However, it doesn't work the other way around; the fact that a strain with a mutation has become more prevalent doesn't mean that the mutation necessarily made the virus more transmissible. Over time, even under what is called "neutral selection" you would expect some mutations to become more prevalent due to chance.
This paper makes the case that there is no evidence of increased transmission from various COVID-19 mutations:
https://www.nature.com/articles/s41467-020-19818-2
On D641G specifically:
Quote:A much discussed mutation in the context of demographic confounding is D614G (nucleotide position 23,403), a non-synonymous change in the SARS-CoV-2 Spike protein. Korber et al. suggested that D614G increases transmissibility but with no measurable effect on patient infection outcome21. Other studies have suggested associations with increased infectivity in vitro18,40 and antigenicity41. Here we conversely find that D614G does not associate with significantly increased viral transmission (median log10(RoHO) = 0, paired t test p = 0.28; Supplementary Data 4), in line with our results for all other tested recurrent mutations. Though clearly, different choices of methodology may lead to different conclusions. A recent study on a sample of 25,000 whole-genome sequences exclusively from the UK used different approaches to investigate D614G. Not all analyses found a conclusive signal for D614G, and effects on transmission, when detected, appeared relatively moderate39.
These apparently contrasting results for D614G should be considered carefully. What is, however, indisputable is that D614G emerged early in the pandemic and is now found at high frequency globally, with 36,347 assemblies in our data set (77.8%) carrying the derived allele (Fig. 1a and Supplementary Data 3). However, D614G is also in linkage disequilibrium (LD) with three other derived mutations (nucleotide positions 241, 3037, and 14,408) that have experienced highly similar expansions, as 98.9% of accessions with D614G also carry these derived alleles (35,954/36,347). It should be noted that the D614G mutation displays only five independent emergences that qualify for inclusion in our analyses (fewer than the other three sites it is associated with). While this limits our power to detect a statistically significant association with transmissibility, the low number of independent emergences suggests to us that the abundance of D614G is more probably a demographic artefact: D614G went up in frequency as the SARS-CoV-2 population expanded, largely due to a founder effect originating from one of the deepest branches in the global phylogeny, rather than being a driver of transmission itself.
If I can translate that, they don't think that D614G is more transmissible than other mutations (but also can't prove the negative).
BC