Lesson 9

Question 1

Invention of Race as Social Construct

Answer 1

Used as a classifier by Portuguese slave traders in 1400s; Enlightenment Era advanced distinction to provide justification for Europeans to colonize Africa; used as a weapon in the Americas

Question 2

Race vs. Ancestry

Answer 2

Race is social construct, ancestry is what actually makes up genetic composition differences; 99.9% the same

Question 3

Human Migration Patterns

Answer 3

Modern humans are 200-300K y.o. Left N. Africa 75K y.a., last place reached was the Americas 15K y.a.

Question 4

Global Genetic Ancestral Variation

Answer 4

Most genetic variation that is unique to the continent within the 0.1% present in Africa, almost 1/3rd of variation is unique to population or continent. Americas are the smallest, with majority of genetic diversity shared between all contents.

Question 5

Historical African Migration Patterns

Answer 5

West African genome is still highly present and distinct in the US due to transatlantic slave trade, hasn’t had gradual spread / intermixing of variation as with nomadic migration patterns.

Question 6

Haplotypes

Answer 6

SNPs that are close together and inherited as an entire block from parent; helps discern ancestry.

Occasionally, SNP mutations such as deletion can predispose certain populations to diseases. However, most genetic variation is in the 98% of the genome that is non-protein coding but rather regulatory. Rare differences in 2% of coding DNA is easier to identify and intervene, leads to discovery gap between common and rare genotypes that are difficult to identify and less accessible to intervention despite being responsible for most diseases.

Question 7

Differences Between Rare and Common Variation

Answer 7

CF (Rare) = mutation on singular gene, easy to model and target

CAD (Common) = additive mutation across entire genome

Question 8

Ancestral Breakdown of Genetic Studies

Answer 8

Despite genetic variation being largest in African populations, 85% of genetic studies have been performed on Europeans –> global implications for relying on predominately euro-centric data.

Question 9

Intrinsic Bias in Machine Learning and Algorithms

Answer 9

When machine learning is coded on initial data including bias, the algorithm will run inputs through under these biased assumptions and skew outputs.

Biased Data –> Biased Model –> Biased Decisions strengthened by feedback loop that trains algorithm.

Question 10

MENA Region

Answer 10

North Africa + Middle East that has been understudied; in census data they are typically considered White / Caucasian which hides their genetic diversity

Question 11

Bias Propagation

Answer 11

Process by which biases inherent in a system, model or dataset are transferred and possible amplified through various stages of data processing, analysis and decision-making

Question 12

Genetic Risk Prediction Accuracy

Answer 12

Up to 75% of negatives may be false negatives in African population when based on European metrics

Question 13

Binary Trait GWAS

Answer 13

Case vs. Control (either has condition or doesn’t)

Question 14

Quantitative Trait

Answer 14

Scales related to having a condition or not, can measure strength of exposures in outcomes etc. (social response, blood sugar, blood pressure, weight etc.)

Question 15

Polygenic Risk Score

Answer 15

Developed using weighted identified risk variants from GWAS studies and applying ML; can associate which SNPs most strongly correlate with the development of diseases.

Question 16

Autism Example: Bias in Machine Learning

Answer 16

In European and mixed ancestry SPARK samples, algorithm trained on European data could significantly deviate between children with autism and unaffected siblings. Could not in those with African ancestry.

Question 17

Accuracy vs. Precision

Answer 17

Precision is internal validity (results are reproducable); accuracy is external validity (results are right)

Question 18

Racism

Answer 18

Process by which systems and policies, actions and attitudes create inequitable opportunities and outcomes for people based on race

Question 19

Bias

Answer 19

The attitudes or stereotypes that affect our understanding, actions and decisions unconsciously

Question 20

Prejudice

Answer 20

Judging, forming and/or acting on an opinion before having all relevant facts

Question 21

Equality

Answer 21

Treating everyone the same, regardless of their differences or needs

Question 22

Equity

Answer 22

Acknowledging that people start from different places and need different resources to achieve equal outcomes

Question 23

Disparity

Answer 23

Differences in health among population groups

Question 24

Inequity

Answer 24

Health disparities that are deemed unfair or stemming from some form of injustice

Question 25

Intent

Answer 25

Mental objective behind an action

Question 26

Imapct

Answer 26

Actual effect or influence of an action on a situation or person

Question 27

Allyship

Answer 27

Active support for the rights of a marginalized group w/o being a member of it

Question 28

Antiracism

Answer 28

Systems and policies, actions and attitudes that oppose racism and promote equity

Question 29

VBAC Algorithm

Answer 29

Predicts the risk posed by trial of labor for someone who has previously undergone cesarean section; predicts lower likelihood of success for anyone identified as African American or Hispanic.