Very interesting open access paper:
If the authors had complete representation in their screens, then they screened about 10 trillion (give or take an order of magnitude) sequences. They identified three sequence families. Let’s call these three sequences (not true, but about as friendly to the ID argument as I can think of), then that is a frequency in sequence space of about 1 in 3 trillion. To make the typing easier, lets round this to 1 in 10^-13.
To make sure we have enough sequences to get at least one of these, we need about 10^15 individual sequences. That is about 50 micrograms, or 1 nanomole, of RNA. If all of this is in, say, 1 liter of soup, then the molar concentration of the RNA mix is 1 nanomolar.
Thus, if aminoacylation can be generalized to all possible RNA catalysts, then a one liter soup of 1 nM RNA should be enough to have a very diverse range of catalysts.
(Someone please check my numbers!)
It would be interesting to see how the fitness landscape looks for different amino acids. As far as I could gather they only tested aminoacylation of different ribozymes with Tyrosine. Presumably the fitness landscape for other amino acids would not be identical, and the degree of overlap and connectivity between the aminoacylating functions of ribozymes for different amino acids would also have some implications for how likely it is that other functions can evolve if one function has already been found.
Are some of the valleys in the Tyrosine aminoacylating ribozyme landscape, peaks in the leucine ribozyme landscape, for example?