I quoted this bit in another comment, but Steven Quay (interviewed by the US gov as mentioned in tfa)

The “engineered” comments refer to common amino acid sequences from lab practices, they leave a signature because ordinary biology is more random.

  The gene sequence for the amino acids in the furin site in CoV-2 uses a very rare set of two codons, three letter words so six letters in a row, that arerarely used individually and have never been seen together in tandem in any coronaviruses in nature. But these same ‘rare in nature’ codons turn out to be the very ones that are always used by scientists in the laboratory when researchers want to add the amino acid arginine, the ones that are found in the furin site. When scientists add a dimer of arginine codons to a coronavirus, they invariably use the word, CGG-CGG, but coronaviruses in nature rarely (<1%) use this codon pair.  For example, in the 580,000 codons of 58 Sarbecoviruses the only CGG pair is CoV-2; none of the other 57 sarbecoviruses have such a pair.