30. Kimura (1968); King and Jukes (1969); Kimura (1983).
31. E.g., Hermisson and Pennings (2005): Fig. 2.
32. Coop, Pickrell, Novembre et al. (2009).
33. This account is taken from Trut (1999).
34. Kettlewell (1955); Kettlewell (1956).
35. Weiner (1994). For a full list of examples of rapid evolution in nonhuman animals, see Winegard, Winegard, and Boutwell (2017): Table 1.
36. Hedges (2000).
37. The wording of the sentence is adapted from Schrider and Kern (2017).
38. Delwiche (2004).
39. Cochran and Harpending (2009).
40. Sabeti, Schaffner, Fry et al. (2006).
41. E.g., Schrider and Kern (2017); Field, Boyle, Telis et al. (2016).
42. E.g., the Singleton Density Score for timing and the Composite of Multiple Signals (CMS) for identifying specific areas within regions. Field, Boyle, Telis et al. (2016); Grossman, Andersen, Shlyakhter et al. (2013).
43. Mardis (2008).
44. Ewen Callaway, “First Draft of Neanderthal Genome Is Unveiled,” New Scientist, February 12, 2009.
45. Olalde and Lalueza-Fox (2015).
46. Akey (2009).
47. “The criteria [for inclusion] were that the study was performed in the HapMap or Perlegen data, lists of all loci deemed as outliers were available as supplemental data, and sufficient information provided information about what genome build was used for the reported map positions.” Akey (2009): 714.
48. Akey (2009): 717.
49. HGDP refers to the Human Genome Diversity Project and CEPH refers to the Centre d’Étude du Polymorphisme Humain.
50. Grossman, Andersen, Shlyakhter et al. (2013).
51. Field, Boyle, Telis et al. (2016).
52. Schrider and Kern (2017).
53. McCoy and Akey (2017): 142.
54. Haasle and Payseur (2016): Table 1.
55. Purifying selection refers to the selective removal of deleterious alleles. Background selection refers to removal of nondeleterious alleles that are linked to alleles that are removed by purifying selection.
56. Key, Fu, Romagné et al. (2016): 8.
57. The literature on methods of identifying SNPs and regions under recent selection pressure is extensive. Recent articles that also contain excellent literature reviews and bibliographies are Alcala and Rosenberg (2016) and Haasle and Payseur (2016).
58. Fossil evidence of another hominin living outside Africa, Homo floresiensis, was discovered on the Indonesian island of Flores in 2003. The evidence that Homo floresiensis is a separate species in genus Homo has accumulated since then, but doubts remain. Perhaps other, still undiscovered species of hominins existed outside Africa. See Detroit, Mijares, Corny et al. (2019).
59. Callaway (2016).
60. Stewart and Stringer (2012). Neanderthal alleles may have increased the risk for certain diseases. It also seems possible that interbreeding with Neanderthals may have reduced human male fertility. Sankararaman, Mallick, Dannemann et al. (2014).
61. Dannemann, Prüfer, and Kelso (2017): 1 of 11.
62. Jensen, Payseur, Stephan et al. (2018).
63. Kern and Hahn (2018): 1366.
64. Voight, Kudaravalli, Wen et al. (2006).
65. “In principle, sharing of signals between populations might also be due to haplotypes that are inherited from the ancestral populations. However, this is probably a small effect since such unusually long haplotypes would be unlikely to survive the effects of recombination for >1,000 generations, separately in each population. Instead, the data suggest that most of the selective events that we detect are local to a single population, but that a significant fraction of the selective events are experienced by more than one population.” Voight, Kudaravalli, Wen et al. (2006): 452.
66. Voight, Kudaravalli, Wen et al. (2006): 451.
67. Akey (2009): 716. See also Ronald and Akey (2005).
68. Herráez, Bauchet, Tang et al. (2009): Table S2.
69. Herráez, Bauchet, Tang et al. (2009): Table S1.
70. Also in 2009, a team comprised primarily of Stanford geneticists analyzed the same database as Herráez, Bauchet, Tang et al. (2009) using still another method of identifying genetic regions subject to selection pressure. They found the same geographic pattern—moderate overlap among European, Middle Eastern, and South Asian populations; little overlap between those three regions and East Asia; and almost no overlap between African and non-African populations. Pickrell, Coop, Novembre et al. (2009).
71. For this figure, the unit of analysis was the population, and the cell entry was the proportion of genetic regions under selection shared by the two populations for that cell. For the visually similar figure in chapter 7, the unit of analysis was the individual and the cell entries were measures of genetic distance—Wright’s fixation index, FST.
9: The Landscape of Ancestral Population Differences
1. Responsibility for the GWAS Catalog was subsequently shared with the European Bioinformatics Institute (EBI). The GWAS Catalog is downloadable free of charge at its website, ebi.ac.uk/gwas. The level of statistical significance required for entry in the GWAS Catalog is p <1.0×10–5, which is more inclusive than the standard for statistical significance in the published literature (p <1.0×10–8). To be eligible for the database, the study must meet certain technical criteria and have been published in an English-language journal.
2. These numbers and those in the figure include all unique SNPs, excluding other kinds of variants in the GWAS Catalog.
3. Lander and Schork (1994), Thomas and Witte (2002).
4. Marchini, Cardon, Phillips et al. (2004): 516. See also Freedman, Reich, Penney et al. (2004).
5. Scutari, Mackay, and Balding (2016).
6. One of the first systematic evaluations of population stratification using polygenic scores (first author was Alicia Martin) calculated polygenic scores for eight well-studied phenotypes and concluded that polygenic scores based on a single-ancestry population have numerous problems. For example, polygenic scores based on a European sample predict that Africans are shorter than Europeans, which is not true except for a few pygmy populations. Polygenic scores for risk of schizophrenia show a lower score for Africans than for Europeans, whereas the actual incidence rates are similar. Alicia Martin (2017): 645.
Since the Martin study, Reisberg, Iljasenko, Läll et al. (2017) and Luo, Li, Wang et al. (2018) have found other traits that are differentially predicted for different populations. Kerminen, Martin, Koskela et al. (2018) and Berg, Harpak, Sinnott-Armstrong et al. (2018) have demonstrated such differences not just across continental populations but within subpopulations of Finns and British respectively.
Duncan, Shen, Gelaye et al. (2018), a systematic review of population differences in the predictiveness of polygenic scores, identified 29 studies that lent themselves to comparisons of a European population with another. Overall, the Duncan study found that the median effect size of polygenic scores was lower in non-European populations. Median effect sizes were 93 percent of that of a matched European sample for East Asians, 80 percent for South Asians, and 36 percent for Africans. The greater attenuation of predictive performance was “consistent with, on average, greater genetic distance between European and African ancestry populations, than between European and other ancestry populations.” (6 of 21). The median effect sizes for the South Asian and African comparisons were consistently smaller than those for the matched European sample. Results for the European/East Asian comparison were intriguingly inconsistent: in six out of the thirteen studies, the median effect size was larger for the East Asians than for the matched European population (fig. 2).
7. Duncan, Shen, Gelaye et al. (2018): 9–10 of 21. Regarding bias in polygenic scoring methods, the authors wrote: “Specifically, linkage disequilibrium (LD) structure and variant frequency are captured imperfectly with current methods (including genotyping and imputation), and they vary across populations, and current
ly available data resources are unequally representative of diverse global populations.”
8. Hannah Devlin, “Genetics Research ‘Biased Towards Studying White Europeans,’” Guardian, October 8, 2018. Examples of earlier murmurings are Bustamante, Burchard, and De la Vega (2011); Lindsey Konkel, “The Racial Discrimination Embedded in Modern Medicine,” Newsweek, October 20, 2015; and Denise Grady, “Genetic Tests for a Heart Disorder Mistakenly Find Blacks at Risk,” New York Times, August 17, 2016.
9. Sirugo, Williams, and Tishkoff (2019): 30.
10. I used a random number generator to order the 962 SNPs and chose the first 500 contiguous pairs that produced correlations matching the correlations for the full sample of 962 to the second decimal place.
11. African: Nigerians, Kenyans, and African Americans. East Asian: Japanese and two samples of Chinese. European: British, Finns, Italians, and European Americans.
12. Author’s analysis based on the GWAS Catalog as of May 2019.
13. The GWAS Catalog associates each SNP with a trait (the variable name is diseasetrait) based on the description of the trait in the journal article in question. The catalog uses the same label for traits studied in different journal articles if the measure is exactly the same. If there is any meaningful difference, the labels differ as well. For example, three labels for risk-taking are “General risk tolerance,” “Risk-taking tendency,” and “Self-reported risk-taking behavior.” Limiting the analysis of broad types of traits to studies that reported at least 100 unique SNPs under the same label in the GWAS Catalog served as a screen for recent studies (and therefore usually more sophisticated ones) with very large samples and required no judgment calls about whether two similar studies should be aggregated.
Major diseases (combined sample = 3,718). The diseases were atrial fibrillation, breast cancer, chronic lymphocytic leukemia, colorectal cancer, coronary artery disease, Crohn’s disease, diisocyanate-induced asthma, diverticular disease, inflammatory bowel disease, lung cancer, multiple sclerosis, Parkinson’s disease, prostate cancer, rheumatoid arthritis, systematic lupus erythematosus, type 2 diabetes, and ulcerative colitis.
Physiological biomarkers (combined sample = 5,298 unique SNPs). The traits were total cholesterol, total HDL cholesterol, LDL cholesterol, triglycerides, BMI, waist circumference adjusted for BMI, waist-to-hip ratio adjusted for BMI, diastolic blood pressure, systolic blood pressure, pulse pressure, intraocular pressure, hand grip strength, height, weight, total body bone mineral density, heel bone mineral density, and menarche (age at onset).
Blood parameters (combined sample = 4,415). Blood parameters were blood metabolite levels, eosinophil percentage of white cells, granulocyte percentage of myeloid white cells, hematocrit, hemoglobin concentration, mean corpuscular volume, mean corpuscular hemoglobin, mean platelet volume, monocyte count, monocyte percentage of white cells, myeloid white cell count, platelet count, platelet distribution width, plateletcrit, red blood cell count, and white blood cell count. This list does not exhaust all the blood parameters that are associated with at least 100 unique SNPs in the GWAS Catalog as of May 2019, but they are representative of the entire inventory.
Cognitive disorders (combined sample = 2,594). Cognitive disorders were autism spectrum disorder or schizophrenia, bipolar disorder, depression, depressive symptoms, major depressive disorder, neuroticism, schizophrenia, and “worry.”
Mental abilities (combined sample = 5,715). Mental abilities were cognitive performance, educational attainment, general cognitive ability, highest math class taken, intelligence, and self-reported math ability.
Personality features (combined sample = 1,319). Personality features were adventurousness, alcohol consumption (drinks per week), general risk tolerance, risk-taking tolerance, life satisfaction, positive affect, subjective well-being, and well-being spectrum.
14. The choice of the particular traits used for this illustrative table doesn’t make much difference. If all of the SNPs in the GWAS Catalog are used, 32 percent of the physiological traits and 34 percent of the cognitive traits have target allele differences that qualify as “large.”
15. The following traits consisted of the SNPs under the same label in the GWAS Catalog: Adventurousness, Positive affect, Life satisfaction, and Schizophrenia. The other traits combined SNPs from more than one label in the GWAS Catalog. Those labels are as follow:
Cognitive ability: Cognitive performance, Cognitive performance (MTAG), Extremely high intelligence, General cognitive ability, Intelligence, Intelligence (MTAG).
Highest math course: Highest math class taken, Highest math class taken (MTAG).
Educational attainment: Educational attainment, Educational attainment (MTAG), Educational attainment (college completion), Educational attainment (years of education).
Self-reported math ability: Self-reported math ability, Self-reported math ability (MTAG).
Neurocognitive function: Cognitive function, Episodic memory, Information processing speed, Logical memory (delayed recall), Logical memory (delayed recall) in mild cognitive impairment, Logical memory (delayed recall) in normal cognition, Logical memory (immediate recall) in mild cognitive impairment, Logical memory (immediate recall) in normal cognition, Reading and spelling, Verbal declarative memory, Verbal memory, Verbal memory performance (immediate recall change), Verbal memory performance (immediate recall level), Verbal memory performance (residualized delayed recall change), Verbal memory performance (residualized delayed recall level), Visual memory, Word reading, Working memory.
Depression: Depressed affect, Depression, Depression (quantitative trait), Depressive symptoms, Depressive symptoms (MTAG), Depressive symptoms (SSRI exposure interaction), Depressive symptoms (stressful life events interaction), Major depressive disorder, Major depressive disorder (broad), Major depressive disorder (probable), Current major depressive disorder.
Neuroticism: Neuroticism, Neuroticism (MTAG).
Worry: Feeling worry, Worry, Worry too long after an embarrassing experience.
Risk tolerance: General risk tolerance (MTAG), Risk-taking tendency (4-domain principal component model).
Well-being: Eudaimonic well-being, Hedonic well-being, Subjective well-being, Subjective well-being (MTAG), Well-being spectrum (multivariate analysis).
Autism: Autism, Autism spectrum disorder, Obsessive-compulsive disorder or autistic spectrum disorder, Social autistic-like traits.
ADHD: Attention function, Attention function in attention deficit hyperactive disorder, Attention deficit hyperactivity disorder, Attention deficit hyperactivity disorder (combined symptoms), Attention deficit hyperactivity disorder (hyperactivity-impulsivity symptoms), Attention deficit hyperactivity disorder (inattention symptoms), Attention deficit hyperactivity disorder (time to onset), Attention deficit hyperactivity disorder and conduct disorder, Attention deficit hyperactivity disorder motor coordination, Attention deficit hyperactivity disorder symptom score, Attention deficit hyperactivity disorder (maternal expressed emotions interaction).
Bipolar disorder: Binge-eating behavior and bipolar disorder, Binge-eating behavior in bipolar disorder, Bipolar I disorder, Bipolar disorder, Bipolar disorder (age of onset <21) or attention deficit hyperactivity disorder, Bipolar disorder (early onset), Bipolar disorder (mania), Bipolar disorder and eating disorder, Bipolar disorder and schizophrenia, Bipolar disorder or attention deficit hyperactivity disorder, Bipolar disorder with mood-incongruent psychosis, Eating disorder in bipolar disorder.
Conduct disorder: Aggressiveness in attention deficit hyperactivity disorder, Anger, Behavioral disinhibition (generation interaction), Callous-unemotional behavior, Childhood and early adolescence aggressive behavior, Conduct disorder, Conduct disorder (maternal expressed emotions interaction), Conduct disorder (symptom count), Early childhood aggressive behavior, Middle childhood and early adolescence aggressive behavior, Non-substance related behavioral disinhibition, Oppositional defiant disorder dimensions in attention-deficity hyperactivity disorder.
>
Alcohol consumption: Alcohol consumption, Alcohol consumption (drinkers vs non-drinkers), Alcohol consumption (drinks per week), Alcohol consumption (heavy vs. light/non-drinkers), Alcohol consumption (transferrin glycosylation), Alcohol consumption in current drinkers, Alcohol consumption over the past year.
Alcohol dependence: Alcohol dependence, Alcohol dependence (age at onset), Alcohol dependence or chronic alcoholic pancreatitis or alcohol-related liver cirrhosis, Alcohol dependence symptom count, Alcohol use disorder (consumption score), Alcohol use disorder (dependence and problematic use scores), Alcohol use disorder (total score), Alcoholism (12-month weekly alcohol consumption score), Alcoholism (alcohol dependence factor score), Alcoholism (alcohol use disorder factor score), Alcoholism (heaviness of drinking).
Brain volumes: Brain connectivity, Brain structure, Brain structure (hippocampal volume), Brain structure (temporal lobe volume), Brain volume in infants (cerebrospinal fluid), Brain volume in infants (gray matter), Brain volume in infants (intracranial brain volume), Brain volume in infants (white matter), Cortical thickness, Dentate gyrus granule cell layer volume, Dentate gyrus molecular layer volume, Heschl’s gyrus morphology, Hippocampal atrophy, Hippocampal fissure volume, Hippocampal sclerosis, Hippocampal subfield CA1 volume, Hippocampal subfield CA1 volume (corrected for total hippocampal volume), Hippocampal subfield CA3 volume, Hippocampal subfield CA4 volume, Hippocampal tail volume, Hippocampal tail volume (corrected for total hippocampal volume), Hippocampal volume, Hippocampal volume in mild cognitive impairment, Hippocampal volume in normal cognition, Intracranial volume, Maximum cranial length, Maximum cranial width, Mesial temporal lobe epilepsy with hippocampal sclerosis, Presubiculum volume (corrected for total hippocampal volume), Subcortical brain region volumes, Subiculum volume (corrected for total hippocampal volume), Superior frontal gyrus gray matter volume, Total hippocampal volume.
Cerebrospinal fluid: Cerebrospinal AB1-42 levels in mild cognitive impairment, Cerebrospinal AB1-42 levels in normal cognition, Cerebrospinal p-tau181p levels, Cerebrospinal t-tau levels, Cerebrospinal fluid AB1-42 levels, Cerebrospinal fluid beta-site APP cleaving enzyme levels, Cerebrospinal fluid biomarker levels, Cerebrospinal fluid levels of Alzheimer’s disease–related proteins, Cerebrospinal fluid p-tau181p:AB1-42 ratio, Cerebrospinal fluid p-tau levels, Cerebrospinal fluid p-tau levels in mild cognitive impairment, Cerebrospinal fluid p-tau levels in normal cognition, Cerebrospinal fluid t-tau levels, Cerebrospinal fluid t-tau levels in mild cognitive impairment, Cerebrospinal fluid t-tau levels in normal cognition, Cerebrospinal fluid t-tau:AB1-42 ratio, Cerebrospinal fluid α-synuclein levels.
Human Diversity Page 53
