It's interesting, that we're getting down to the sizes of self-replicating RNA that realistically can form by a complete accident.
Getting this sequence by random chance out of a pile of nucleotides is a 1 in 2^90 chance. That's around 1.2*10^27 or just around 20000 moles! Not at all an impossible number.
2009 paper: Self-sustained replication of an RNA enzyme [0].
See also: Spiegleman's Monster
[dead]
Quinozyme