January 21, 2016 at 05:28AM

Today I Learned: Today's TIL comes from a lecture by David Baltimore, one of the rather more prestigious names I've had the privilege of listening to. The lecture was on human gene editing, and unfortunately for me it was a lecture for general scientists (mostly Caltech undergrads), so I knew most of it. There were still some nice tidbits, though. 1) About 15 base pairs of sequence is sufficient to be unique in the human genome. Either that's a very rounded number, or that's surprisingly low -- there are something like 6 billion bases in the human genome, so you would expect log_2(6 billion) ~ 32.5 bits ~ 16 bases to be unique. 2) Zinc finger proteins (ZFs) and TALEs are both classes of protein that recognize and bind to specific DNA sequences, and both can be modified to cut, much like Cas9. Both proteins are highly modular, with each segment binding to another little bit of DNA. ZFs and TALEs can be redesigned to target different, arbitrary sequences by rearranging, adding, and removing segments, but the process is rather arduous. Anyway, in this lecture, I learned that ZFs a) were discovered first, and b) have segments that target several bases at once, so the library of ZF subunits is on the order of dozens of variants, and c) are proprietary in their recombinant form, or at least were up until recently?. TALEs, on the other hand, a) were discovered after ZFs, b) are native to plants, c) have subunits that recognize individual nucleotides, so there are only somewhere around 4 variants, and d) are in the public domain. Basically, TALEs are strictly better for a bioengineer. 3) "Most" Mendelian diseases (that is, diseases caused by a single DNA mutation) are dominant. That surprised me a lot, and I'm actually a little suspicious of this -- how could gain-of-function mutations with negative side-effects be more common than loss-of-function mutations with negative side effects? In general, breaking stuff is much easier to do by mutation than making new stuff, even bad stuff.