February 23, 2016 at 01:02AM

Today I Learned: 1) Flu kills about half a million people a year. Half a million! That's... surprisingly large. Also, fever is a really good marker of having the flu -- if you have a fever and a cough during flu season, odds are around 80% that you have the flu. 2) Spiegelman's Monster is GGGGGCCUUCCUUGCCAGACCGCCCCAAUACAGUAACAGCGCGUUGGC. It is the smallest known RNA sequence capable of replication with only an enzyme (specifically, the RNA polymerase of bacteriophage Qβ, which is unusual for replicating RNA from and RNA template). It was derived (by Spiegelman) from the sequence of Qβ, which is an RNA virus, by putting a bunch of Qβ RNA genome in a test tube with Qβ RNA polymerase for a while. The genome replicated a whole bunch, occasionally with mutations. Shorter variants replicate faster in such an environment, so there was selection for short mutants. In the 1965 paper, the shortest sequence the researchers got was 218 bases, after a mere 74 generations of replication (I can't give you the sequence because it wasn't sequenced -- this was 1965, after all). A later replication of the study wen longer and eventually got the sequence I showed above. This was one of the classic experiments of minimalist evolution. 3) Ribozymes (RNA molecules that function as some kind of enzyme) may be much more common than I thought in sequence space. A 1993 experiment by Bartel and Szostak tried discovering new RNA ligases from random 220-base-pair RNA sequences. They built a library of about a quadrillion (10^15) random RNA sequences, then selected for sequences that could catalyze a particular ligation reaction. After four rounds of in vitro selection (without mutation) they eventually isolated *68* distinct RNAs capable of ligating about 10,000 times faster than with no catalyst. That's not particularly good, for an enzyme (it's terrible) but it's quite good for an RNA, especially a random one. With error-prone PCR and a bunch more selection, they were able to get ligases with activities of several million times the background rate. Assuming the random RNAs were actually representative (which I'm actually pretty willing to grant in this one), that implies that about one in every 15 trillion (~10^13) RNAs is at least a semi-functional ligase. That's really impres sive. Given that there are 4^220 ~ 10^132 possible RNAs of length 220 nucleotides, that implies that there are somewhere around 10^119 RNA ligases somewhere in sequence space. That's a kind of number big enough to start comparing to the estimated number of "molecular events" that have happened in the entire unvierse over the entire history of the universe... and that's just ligases! Think of what else RNA can probably do!