What are direct-to-consumer genotyping companies?
- Customer provides cheek swab or spit sample
- Company genotypes customer using a SNP array e.g. the Illumina or Affymetrix chips used in HapMap, 1000 Genomes, etc. type projects
- Company provides a report on ancestry, possible relatives or health-related topic
- Company sometimes uses (anonymised) data for research
- Customer also obtains data as text file – they can choose to share this (anonymously or otherwise)
Uptake of Direct-To-Consumer companies
2020: 23andMe report 12 million people tested; Ancestry DNA even more customers
23andMe
- Founded in 2006
- 12 million customers
- Sometimes difficult relationship with US FDA (Food and Drug Administration)
- Valued at $US3.5 billion
- Initial aim- tell people if they were more likely to get genetic disorders and reduce the risk in that through lifestyle changes - data given is accurate, the interpretation isn’t accurate
How do Health Tests work? (literature based)
- Company (e.g. 23andMe) continue to monitor results of published GWAS studies on all diseases, traits, etc
- They use the same SNP chips used in most GWAS studies- relies on the information the customer provides in the first place
- They calculate an individual’s risk/carrier status etc, based on the customer’s genotype at the significant GWAS SNPs
Own research
- Customers complete questionnaires on variety of traits
- As database grows (to become bigger than published studies), 23andMe perform their own GWAS
How do health tests work? (own research)
- Customers complete questionnaires on variety of traits
- As database grows (to become bigger than published studies), 23andMe perform their own GWAS
23andMe Health Tests
Health and ancestry service - and risk of getting genetic disorders
Example of 23&Me data being used in research
Now a massive data set- tends to use the same SNP chips - have collaborated to increase sample size - trying to find alleles associated with traits
Ancestry
- Company compares your SNP genotypes to those of reference populations
- They try to predict the populations your ancestors came from (methods not usually published)
- Marketed at people interested in family trees, genealogy, etc
-Large customer base of people sharing family trees, extended to DNA testing
AncestryDNA- example report
Example report
This person has ancestry from Western Europe and Mexican native Americans…
… and small proportions of ancestry from many other places?
What does chromosome painting show
shows which genomic regions come from which populations
Genetic astrology?
The same person can get different reports from different companies…..or even from the same company!
Depends on:
- Algorithm used
- Reference populations used
- How reference populations are labelled
- shouldn’t interpret these tests too literally
- Reflecting migratory events that have happened thousands of years ago - not telling us about our recent past and for example not where our grandparents are from
What is pedigree collapse
Our ancestry links back to the same person via many routes. Therefore, we have far fewer ancestors than 2n when we go back n generations
Whats the modelling approach for pedigree collapse
- Predicts that most recent common ancestor of all living humans lived just ~3500 years ago
- Going back a bit further (~7500 years ago), everybody alive was either an ancestor of all living humans or no living humans- ancestry comes from relatively few people due to only a number of finite possibilities
But most of our ancestors have not contributed to our genome- what does this mean?
- During meiosis, there are an average of just 33 recombinations across all 23 chromosomes
- We inherit large blocks of chromosome together
- In this cartoon, the green, pink and grey great-grandparental chromosomes have not been passed to the granddaughter
- Has some parts from the grandparents that are completely missing - so can have ancestors that haven’t contributed anything to your genome
Identifying relatives through DNA matches
- If databases are large enough, there is a high chance that somebody submitting a sample can find distant relatives e.g. 3rd or 4th cousins
- This is why AncestryDNA appeals to family tree enthusiasts
What was the golden state killer
The ‘Golden State Killer’
- Between 1974-1986 at least 13 murders, >50 rapes, >100 burglaries during 3 separate sprees across different parts of California
- No clear suspect
- In ~2002 crime scene DNA compared to prisoners on Death Row, in case perpetrator has subsequently been arrested for other crimes. No match
- In Jan 2018 crime scene sample was submitted to GEDMatch
What is Gedmatch
- GEDmatch is free software where DTC customers can upload their genotypes to try and identify relatives
- Killer’s DNA had no close relatives but 10-20 possible 3rd or 4th cousins on GEDmatch
Using genealogy to trace the killer
- Built a large family tree (using birth records, marriage certificates, death records etc), that linked the third cousins to all of their relatives
- Two people were identified that could be the killer – one of them (deceased) ruled out when a close family member provided DNA that was unrelated to crime scene sample
- April 2018: Other suspect - Police surreptitiously took swab from car door handle and from tissue in a bin. Perfect match to crime scene sample
- April 24th 2018 – police arrested Joseph James D’Angelo (then aged 72). Former policeman who lived in the areas where the crimes took place. Court case expected to last 10 years.
- March 2020 – offered to plead guilty if death penalty threat removed. State prosecutors declined.
- August 2020 – life sentence with no parole, as part of plea bargain. Admitted to additional crimes.
Solving cold cases via distant relatives
- Since Golden State Killer, same technique used in other historical serious crime cases
- SNP chips mean we can distinguish between non-relatives and distant relatives
- Therefore, criminals can be identified because distant relatives, who they have probably never met, have uploaded their data to databases like GEDmatch
- Power to identify criminals will increase as databases grow
- Not all companies allow their data to be used in this way
