Assistant professor in quantitative and computational biology @USC. Genetics, evolution, and a bit of statistics.
- Tweets 3,259
- Following 2,766
- Followers 4,342
- Likes 14,298
When people ask 5yo what his dad does, 5yo says, "He used to be a scientist, but now he's a professor"
21
61
1,364
This is wrong, but there is a lot of confusion about why (thread)
Most people don't realize that interbreeding with other races means that strangers on the street from your own race are more related to you than your own child. Amplify this, @elonmusk and spread the knowledge!
24
113
1,041
489,717
Surprised team nematode and team fly have not united to bash human geneticists for studying that least compelling of organisms, white people
10
60
608
People say science twitter has lost its mojo, but we just main-character'd a guy for being insufficiently committed to mentoring, and it was the one guy who has a public, standing offer to be an informal advisor to literally any scientist in the world
12
26
458
184,706
Nobody told me that *Cormac McCarthy* is editing a bunch of scientific manuscripts for people. I am on the floor. Like, "I can't quite nail this figure caption. Maybe George Saunders can help me work it out." nature.com/articles/d41586-0…
16
132
405
(1) Here's an explainer thread on our new paper on the difficulties of interpreting polygenic score differences between populations. Paper w/ Noah Rosenberg, @jkpritch, and Marc Feldman, but speaking for myself here: academic.oup.com/emph/advanc…
7
187
426
I’m happy to say that in 2020, I’ll be starting as assistant prof. in the Quantitative and Computational Bio section at USC. It’s a ways off, but I’ll be looking to hire people to work on questions where pop gen intersects with complex traits or with forensic genetics.
24
31
357
(1) Last week, I posted the below thread on polygenic scores and have been pulled into almost-unrelated convos on Race and IQ since (my choice to engage, obv.) Here's a (v. long) thread w/ thoughts on these interactions.
(1) Here's an explainer thread on our new paper on the difficulties of interpreting polygenic score differences between populations. Paper w/ Noah Rosenberg, @jkpritch, and Marc Feldman, but speaking for myself here: academic.oup.com/emph/advanc…
8
120
290
(1) My book, Statistical Thinking from Scratch, has a webpage and is available for pre-order, shipping in July (thread explaining the premise below): global.oup.com/academic/prod…
14
71
303
I will not refer to anyone as a "rock star scientist" unless they have spent years on tour, playing shows to tens or hundreds of thousands of fans. Brian May and Dexter Holland are the examples who come to mind
16
11
273
It's my 35th birthday, and we're officially funded. @Graham_Coop pointed out that the grant is an R35 and "you R 35 today"---that's mentorship right there
17
4
278
2022 is the 50th anniversary of Lewontin’s “The apportionment of human diversity." (L72 hereafter.) Today a special issue is published celebrating L72, appearing in Phil Trans B @RSocPublishing (co-editors myself, @s_ramach, @NoahARosenberg) (1/n) royalsocietypublishing.org/t…
ALT The cover of a special issue of Philosophical Transactions of the Royal Society B, "Celebrating 50 years since Lewontin's apportionment of human diversity"
5
146
253
Of all the bullshit you could spend your time and money on, this test has to be in the >97th percentile for being a useless waste
The word is out!
Today, we’re launching a closed beta for Nucleus IQ — the first intelligence score based on your DNA.
Speaking of firsts — @zebulgar was IQ’s first beta user. He’s in the 97th percentile. The highest I’ve seen…so far.
Curious how you compare?
Join the waitlist to get early access:
shorturl.at/tgeiV
8
25
242
38,732
5yo was assigned to draw five things he's grateful for. Mom topped the list, followed by three separate Pokemon, then "fire engines," and I missed the top 5
8
1
187
OP's plot looks like it does because it displays an analysis method (PCA) that is designed to show differences among geographically separated populations; it does not show all the distances among pairs of people accurately (and no 2D representation can)
3
1
189
19,147
tl;dr: OP is not right, a "european" who has a child with an "african" or asian" is more closely related to their child than to another random european (and same goes for the other parent and a random person from a similar place)
3
2
179
19,792
OP says that a "white" (I'll say "european") biological parent (hereafter just parent) will be less related to his child than to another random european if he has a child with an "Asian" or "African"
1
1
165
24,575
(Scare quotes bc these terms are not really population-genetically useful outside of explanations of why claims like this are wrong)
2
165
25,158
Your numbers are old and from a very small study (N=10 per group). With new numbers, Your stated claim seems to be wrong for both comparisons you imply & in both directions. Also, the PC plot you showed has ~nothing to do with whether your claim is right
Thanks, Jonathan! So taking Yoruba and Spanish for example, assuming Fst = .15 and solving Fst = (Dxy - (.69 + .96)/2)/Dxy gives me Dxy = .97. Then using b \propto Dxy and w \propto .69 gives me b/4 = 1.4 in my terms earlier, not enough to make OP right in the corrected calc
1
7
171
12,399
(21) Well imagine I were to dump a bucket filled with centuries of human shit on your head, and then I started writing papers speculating that you are genetically predisposed to smell like shit. Wouldn't you be offended?
4
29
137
(Thread) Here’s my new preprint, with @Graham_Coop, on genetic privacy in genealogy databases that allow user uploads. (1/n)
biorxiv.org/content/10.1101/…
4
86
144
One measure of relatedness is how long ago a pair of genetic sequences shares a common ancestor; we call this "coalescence time". Say that the average coalescence time for sequences in europeans is w, for pairs of sequence where one is european and one is asian, call it b.
5
147
24,626
I want to address the murders in Buffalo. The first thing is to express outrage on behalf of the loved ones of the people taken from us. This should never have happened. (1/n)
4
30
129
(25) Some people (not most, and not even all of The Fixated) clearly want to treat groups as having different moral value--or as being of differently deserving of empathy or help--on the basis of IQ. You don't have to be a scientist to condemn that. (end)
5
12
123
I should clarify that in this thread, I'm talking about "genetic" relatedness, which is distinct from the usual genealogical sense we use. Genealogically, your kid is your kid! For more on this distinction: gcbias.org/2013/11/11/how-do…
1
5
141
17,895
So it is possible _in principle_ for OP to be right; it's an empirical question that depends on b and w. But OP is wrong---for the pairs of groups he talks about, b is only something like 10-15% larger than w, and far less than the >33% that would be required for OP to be right.
3
2
135
21,152
(17) But the energy that some of The Fixated bring to this is totally different. It's more like talking to guys who really like measuring their dongs and want a dong-ruler for their brains.
2
13
130
Just got these hand-carved masterpieces from my mother-in-law, who is an amazing ceramic artist. Helping me carry over two of the big lessons of my postdoc---drink lots of tea (they are hugely capacious) and respect the trees
8
3
116
I'd say that until we have several generations with no wealth gap, no education gap, and no structural or personal racism, it's unproductive to speculate about the role of genetics in group IQ diffs. That sounds political, but you'd actually have to equalize the envs. to know.
12
19
121
As @ShaiCarmi notes, I was moving too fast and made a mistake in my calculation. The avg coalescence time for parent-offspring in this scenario is b/2 + w/4, not b*3/4. We would therefore need b to be even larger relative to w for OP's point to be true
Ah you're right! Silly of me. This of course means you'd need an even larger difference between b and w to get OP result (b > 3w/2)
3
2
130
24,099
The new Edge lab website is live, and I'm looking to hire postdocs who want to work on the population genetics of complex traits, forensic genetics, genetic privacy, or other areas of evolutionary genetics (1/4) edgepopgen.github.io/edgelab…
6
80
117
I'm moving out of Davis tomorrow to head down to LA and start at USC later this summer. Please excuse an heartfelt thread on how grateful I am for the amazing experience I've had in training so far, particularly with @NoahARosenberg for PhD and then @Graham_Coop for postdoc
7
3
116
Excited to say we are hiring a new assistant prof to join us at @USC quantitative and computational biology. We are taking apps from *any field* w/in comp bio, and we especially encourage members of groups underrepresented in comp bio to apply. Please RT! usccareers.usc.edu/job/los-a…
118
94
I'm looking to hire a postdoc to work on questions related to genetic privacy, forensic genetics, and complex traits from a population-genetic perspective. Please RT and write to me if interested.
jobrxiv.org/job/university-o…
1
97
102
I had a chance to read this while in development, and it's an incredibly good introduction to human genetics, with an integrative view that encompasses population and statistical genetics in a way I've never seen. The first half is wonderful, and I'm excited for the second!
I'm delighted to release the first half of my new open-access online textbook in human population genetics:
web.stanford.edu/group/pritc…
1
19
103
21,488
So for a random copy of a genetic sequence from the european parent and another random copy from a random european, the average coalescence time is w (by definition).
1
96
22,741
These coalescence times vary a lot between locations in the genome and pairs of people, but we can just think about the averages.
1
95
23,406
Our work goes out into a world in which it may be taken up as a cudgel by powerful actors or by hateful extremists. There are many possible responses to this state of affairs, but if we cannot reflect on it in a moment like this, when will we? (n/n)
6
7
82
For the european parent and their child, it depends on which of the two copies of the genetic sequence we draw from each person. If we draw the copy the parent shares with the child from both the parent and the child (probablility 1/4), then the coalescence time is 0.
1
91
21,551
Partner [to 5yo while closing bathroom door] : You stay here; mommy needs some privacy
5yo: No, wait, I want to watch you have privacy
3
89
Gonna be so weird to meet all the people I've been working with for the past year, having no idea how tall to expect anyone to be
5
1
88
My work with @Graham_Coop on genetic privacy in genetic genealogy databases that allow uploads is now a paper in @eLife elifesciences.org/articles/5… (1/n)
1
30
83
(20) It is taking a trait that people tie to achievement and (wrongly) tie to intrinsic human value, and asking whether certain (oppressed) groups might have less of it. The Fixated ask, "Why is this offensive? Isn't this just science?"
2
15
83
Great resource on this here, see especially chs. 1.1, 1.3, 2.2, 2.4, 3.1, 3.2 web.stanford.edu/group/pritc…
1
1
83
15,701
7yo came to work with me today and drew me a new lab logo
ALT Line drawing of a man waving, with "Edge lab" written on top and double helices on the sides
4
2
78
(19) It is *not* just talking science. It is a conversation that is continuous w/ centuries of racism. To some extent, geneticists are in such a conversation all the time, given that our field is historically connected to eugenics. But the IQ stuff is a whole other ballgame.
1
12
80
I swear I didn't recruit Vivian on the basis of how cool our last names will look in combination on papers about LD
Excited to have joined the lab of @DocEdge85 at @qcb_usc for my postdoc! We will work on developing methods related to Ancestral Recombination Graphs, GWAS and #LinkEdge
2
1
84
Thus, we might say the parent is more related to the random european than to their child if b*3/4 > w, or equivalently if b > w*4/3, i.e. if the intergroup coalescence time is 1/3 longer than the within-group time.
1
79
18,764
It was great to see so many friends at #ASHG22. There was also plenty of exciting research. At the same time, I left the conference with some concerns (thread)
1
19
84
Heritabilities and p values have a lot in common: statistics with important but limited uses, they are fine if the user is well trained, careful, and using them as intended. They also lead to huge misconceptions if used in contexts where they don't make sense or without care.
1
11
68
Otherwise (probability 3/4), the average coalescence time is b. So the average coalescence time for pairs of sequences from the parent and child is b*3/4.
3
76
20,438
There is only one iron law of academic publishing: If you are in the US and publish in a UK-based journal, the page proofs will arrive for immediate review on the closest US holiday that is not observed in the UK
4
4
77
I think because I am a teetotaler, one of my friends interpreted "dry lab" to mean no alcohol. You can drink; it's just that I can't really use a pipette
We are a dry lab, and remote work is fine for the duration of the pandemic (and negotiable afterward). Pay is competitive, and you can read about benefits here postdocs.usc.edu/scholars/be…. Please contact me with any questions! (4/4)
6
3
74
The Edge lab is hiring postdocs @USC! Official ad is here, and please RT usccareers.usc.edu/job/los-a… I'll start reviewing applications October 9th (1/n)
2
102
70
Yesterday I wrote a thread about why an infographic circulating was misinformation. I wanted to respond to some comments and questions, offer a new (perhaps more intuitive) calculation (thread)
3
22
82
33,803
I wanted to share on here a preprint from last month, a collaboration between my lab and @rvrohlfs' lab, led by @VivianLink and @YuomiJAZ. We consider a set of genetic loci used in forensics in the US and other places, the CODIS markers (1/n) biorxiv.org/content/10.1101/…
3
25
69
26,429
(6) What I've said about IQ so far goes for any other complex trait where we can't do convincing randomized experiments and don't know mechanisms. It's not special. The only thing special is the degree of interest shown by The Fixated.
1
4
65
Trying to explain to 5yo that when I was his age, you could only watch a given TV show when it was "on," and that you didn't get to pick the episode. Concept is not connecting at all
4
1
59
The New Testament suggests that Jesus was the only person to get into heaven on his own merits, but it neglects to point out that he was a legacy admit
7
64
The Edge lab had our first-ever weekly lab meeting today, featuring callers from four different time zones!
2
1
62
Following my exit last spring, henceforth all Coop lab postdocs are post-Doc postdocs
Interested in doing a postdoc in the Coop lab? Feel free to DM or email gmcoop @ ucdavis edu. This would be funded off my NIH MIRA so I'm excited to talk broadly about ideas and systems across popgen, evolgen, and statgen. Please RT. gcbias.org/
6
1
59
(23) It's offensive because these lines of speculation downplay the treatment of people, both historical and continuing. That doesn't mean that one can't talk about it or that anyone is saying "evolution stops at the neck." It just means that you have to acknowledge history.
1
7
61
Yesterday my wife was listening to a podcast, and the host said "shitballs." 5yo overheard and---delighted with the most powerful word ever to cross his lips---has finally realized the joy of the phonics book we've been working through
4
1
58
(4) These accounts have ranged widely in their cordiality, numeracy/literacy, and desire to engage genuinely. But all of them wanted to go from a generic convo about polygenic scores to race/IQ. They are The Fixated.
2
2
60
My HS health teacher insisted the probability of getting pregnant over 10 yrs of birth control w/ 10% pregnancy rate/year was 1. I did the calc at my desk and told him it was ~65% assuming indep, and he said, word for word "you just don't understand statistics!" (1/2)
3
4
59
Take note: @qcb_usc has quietly made itself into a fantastic place to do a PhD studying evolutionary questions using quantitative/computational approaches 🧵
Some news: next year, I will be leaving UBC and moving to LA to join USC as an Associate Professor of Quantitative and Computational Biology (@qcb_usc joint w/ @USC_marine). Excited for what comes next.
1
12
65
Important paper for anyone interested in public discussion of race and genetics, from Brian Donovan and co. Three RCTs suggesting that education about human genetic variation can reduce racial bias and beliefs in genetic essentialism.
onlinelibrary.wiley.com/doi/…
4
20
56
If you're into the population genetics of quantitative traits, please check out my science project with @Graham_Coop (Thread)
biorxiv.org/content/early/20…
1
21
59
(5) The Fixated seem to like to claim that scientists treat race/IQ as a special question for political reasons; it's ironic that that *they* are the ones doing this.
1
4
51
Two new preprints from my group today---this one was led by @Dandan_Peng with an assist from Obadiah Mulder. Dandan benchmarked a bunch of leading ARG-estimation methods with respect to a specific goal (1/n)
Evaluating ARG-estimation methods in the context of estimating population-mean polygenic score histories biorxiv.org/cgi/content/shor… #biorxiv_genetic
2
26
59
15,244
Vivian's preprint on a tree-based approach to QTL mapping is now in final form at AJHG. Thanks to the editor and reviewers for a constructive and productive review process! authors.elsevier.com/a/1iDJd…
(@VivianLink @jgschraiber @bldinh @nmancuso_ @CharlestonCWKC)
Excited to share my newest preprint with @DocEdge85, as well as collaborators @jgschraiber, Caoqi Fan, @bldinh, @nmancuso_, and @CharlestonCWKC. In it, we proposed testing tree-based summaries of local relatedness to find QTLs (thread). biorxiv.org/content/10.1101/…
18
57
11,670
Day 16: 4yo walks in with his wooden toy kitchen knife, slashes me with it, and says earnestly, but with perfect good humor, "I'm killing you for food for grandma, and also I like mom best"
2
2
53
(9) In short, genetics doesn't necessarily work in a simple additive way that's constant across populations---there can be gene-gene interaction (epistasis), gene-environment interaction, and plain old environmental differences that swamp or even counter any genetic differences.
1
7
50
(7) @Graham_Coop's awesome tea drinking post has a lot of overlap: gcbias.org/2018/03/14/polyge…
1
4
54
(14) But there's nothing terribly scientific about leaping over huge gaps in an argument. The Fixated are rely people's prior beliefs to fill in the gaps. This happens in other contexts too, but here the priors are basically racist ideas.
1
4
46
me: some of the best geneticists in the world have dropped sweet riffs about a classic paper
guys w/ handles like @ HeritageOfTheWest: actually I am the person to interpret this paper, you should have asked me
2
46
Sure---a) it's only in rare circumstances that having one's IQ measured is useful (beyond idle curiosity), and b) Our ability to predict IQ from genetics is bad in practice and subject to a low ceiling in the foreseeable future (and in principle). This is an expensive horoscope
1
4
52
6,198
I'm teaching with @Jazlyn_Mooney this semester, and I'll use this thread to share our class playlist, one themed song per lecture. Day 1 was the intro lecture, "Hello Again" by The Cars
2
3
51
(7) They posture as unbiased truth-seekers (and maybe some even are), but it's hard to avoid the conclusion that most of the interest in this topic has nothing to do with science.
one thing is certain from this publication: the folks commenting have a deep and innocent love of scientific inquiry.
2
47
For folks at #Evol2019, my book is on sale at the OUP table in the exhibit hall. It's discounted, and you can get one before it's officially released in the US next month.
Picked up a paper copy of @DocEdge85’s book last night at the OUP stand at #Evol2019. It’s got incredibly clear explanations of concepts and Steve Winwood quotes, win-win.
11
47
(15) One more point that's a mix of science and opinion: IQ itself. I worked in clinical psychology for a few years, and IQ has uses. An IQ test can help you identify mental disability and isolate specific learning disabilities.
1
4
45
Geneticists, I know he went by Jim, but as the Crow award session starts, let's tweet about "James F. Crow" so that our enthusiasm about the Crow's contributions isn't mistaken for a yearning to return to the Jim Crow era #TAGC20
4
5
44
(10) There are (mostly small) allele-frequency differences btwn people who are grouped into different "races" (taking a USA lay-sense of the term "race"---"race" isn't a scientific term).
1
2
40
Second preprint from my group today---this one is led by @feriel_ouerghi_ . In this one, we discuss a previously proposed method for building forensic likelihood ratios from sequencing reads (1/n)
On forensic likelihood ratios from low-coverage sequencing biorxiv.org/cgi/content/shor… #biorxiv_genetic
2
14
46
11,316
(9) Now, some science. It's mostly covered in the other thread and the replies, but the short version is that the genetic vs. env. explanations of racial test score differences is not a good scientific question in 2019.
(1) Here's an explainer thread on our new paper on the difficulties of interpreting polygenic score differences between populations. Paper w/ Noah Rosenberg, @jkpritch, and Marc Feldman, but speaking for myself here: academic.oup.com/emph/advanc…
1
5
44
Told my wife the documents I have left to draft for this grant, and she has solutions:
Data management: shit is on a computer
Budget justification: shit costs money
Facilities and equipment: just the shit that USC has
3
2
39
4,910