>> Examination of SEQ ID11652 revealed that the match extends beyond the 12-nucleotide insertion to a 19-nucleotide sequence: 5′-CTACGTGCCCGCCGAGGAG-3′ (nt 2733-2751 of SEQ ID11652), such that the resulting mRNA would have 3′- GAUGCACGGGCGGCUCCUC-5′, or equivalently 5′- CU CCU CGG CGG GCA CGU AG-3′ (nucleotides 23547-23565 in the SARS-CoV-2 genome, in which the four bold codons yield PRRA, amino acids 681–684 of its spike protein). This is very rare in the NCBI BLAST database.
I don't like "this is very rare"
How rare?
It's a database. Query it, and confidently give numbers to support your statement.
An n of 1 has infinite variance. The relevant rarity, though is not the database, but rather "in the universe of viral sequences.
The database biases for sequences that we've sequences, so at best you can give a really shitty estimate.
Fwiw I believe the story that this was a stack overflow copy-paste operation from the moderna sequence, but I can only ever call this a strong belief[0], with no numbers behind it, unless someone comes forward and admits having done it.
[0] why strong? Because it follows the scientific method. If the hypothesis is that it's a lab leak, then your prediction is that existing sequences would bleed through. A bit crazy that we only found this now, hell I could have done this blast search years ago, but it is what you would expect to find.
I would say statistically it is not rare. I mean it's a sequence of 12 where each item can only be C,T,A OR G from the little I understand about DNA. It would be quite a bad password even though it's 12 characters long.
statistics in sequence matching depend on underlying base rates; out of the 4*12 (2*24) possible sequences, you will see some never, some many, and many some times.
I don't like "this is very rare"
How rare?
It's a database. Query it, and confidently give numbers to support your statement.