A THREAD about our story today on "Mendelian Randomization". Wow even the name of this thing is daunting. But you'll be hearing more and more about it. cdn.technologyreview.com/v/images/causa…
2/Big thanks to @garytaubes for writing today's story in Technology Review. technologyreview.com/s/611713/resea…
3/Mendelian Randomization named after Gregor Mendel, monk who discovered laws of inheritance. It's a method that uses the "random" way each of us inherits genes as a lever to get at the true effect of life exposures, such as diet, drugs, pollution, education, etc. 
4/ Here's a video that explains it from @mendel_random , the British epidemiologist who has been developing + promoting this technique.
5/ Have you heard the phrase "correlation is not causation"? That's a big problem in public health. Like, is drinking red wine good for you? The way the news headlines go back and forth you know no one has a clue. From our story today.
6/ Imagine then, a technology that can determine which correlations are true causes and, in some cases, distinguish cause from effect.
7/ One you look for it, Mendelian Randomization is EVERYWHERE in the genetics literature. Just this week, @23andMeResearch along with European scientists looked at whether smoking pot causes schizophrenia, or vice versa. gizmodo.com/a-genetic-stud…
8/ It's friggin hard to explain tho, and I won't try in this thread. From our story, an anecdote in which @mendel_random tries to explain it to a colleague.
9/ My favorite mendelian randomization study has got to be the one that asked if wearing glasses turns people into book worms, or vice versa. (turns out the studying causes myopia, overall). telegraph.co.uk/news/2018/06/0…
10/ More seriously, billions $$ at stake when it comes to figuring out what causes heart disease, and therefore what drugs to develop. Like this whopper from 2012 in which @skathire et al found "good" HDL cholesterol doesn't protect against heart disease. thelancet.com/journals/lance…
11/No wonder, then, that all those efforts to create HDL-raising drugs were a flop. What looked like a cause was just a correlation.
12/The author of our story, Gary Taubes, is (in)famous for his role in the diet wars. He's basically of the view that sugar is poison and fat ain't the villain it's made out to be. Here's a review of his book "The Case Against Sugar" nytimes.com/2017/01/02/boo…
13/Small world aside: I was Gary's intern c. 1994 at the beginning of my career. I learned a lot from him. Mostly about how science works. Say, how to look at list of references in a paper. They appear boring. But actually blood sport.
14/Because of Gary's role in the dietary wars, which have been going on for YEARS, Gary became interested in Mendelian Randomization. The technique could determine which side is right!
15/I was interested in Mendelian Randomization for a totally different reason. It's because it's a killer app of "polygenic scores." These scores are the newest way to estimate genetic causes of any trait. It's a tech we've been writing about a lot. technologyreview.com/s/610251/forec…
16/Turns out polygenic scores are pretty much jet fuel for Mendelian Randomization. It's now possible to have a measure of the genetic basis of ANY TRAIT (however poorly) which basically opens up ANY TRAIT to fun and games with Mendelian Randomization. /END
