A new paper has appeared at BioRxiv, The Genomic Health Of Ancient Hominins (2017) by Berence, Cooper and Lachance.
Important results are available at: http://popgen.gatech.edu/ancient-health/.
While the study’s many limitations are obvious to the authors, they still suggest certain interesting possibilities as the most important conclusions:
- In general, Genetic risk scores (GRS) are similar to present-day individuals
- Genomic health seems to be improving over time
- Pastoralists could have been healthier than older and modern populations
Some details and shortcomings of the study (most stated by them, bold is from me) include:
- Allele selection: only some of the known autosomal disease-associated SNPs were included
- Discovered disease-associated SNPs are known to be biased toward European diseases
- Ancient sample selection and genomic quality: only 147 ancient genomes were included, from 449 available, with a conventional cut made at 50% of the focal 3180 disease-associated loci. These samples did not include the same loci. All this can affect whether an individual has high or low GRS (a relationship was found between GRS percentiles and sequencing coverage for ancient samples).
- Phase 3 of the 1000 Genomes Project was used. However, many disease alleles that segregated in the past remain undiscovered – therefore, GRS for ancient individuals should be considered to be underestimated.
- Genetic risk scores were calculated for each individual (with different sets of disease-associated loci), hence they were not comparable across individuals. So GRS were standardized as GRS percentiles, with certain assumptions, comparing them to modern individuals
- Multiple comparisons with all data available, using multiple groups, in the small sample selected: comparisons were made between standardized GRS percentile, sample age (i.e. estimated date), mode of subsistance, and geographic location.
- Older samples have worse coverage, especially Altai Neandertal, Ust’-Ishim, and Denisovan (which might influence results in hunter-gatherers)
- Northern ancient individuals (using latitude values) show healthier genomes: but, most ancient individuals are from Eurasia, and samples are heterogeneous.
- Agriculturalists show a higher genetic risk for dental/periodontal diseases than hunger-gatherers and pastoralists. However, this disease has the smallest number of risk loci (k = 40), so risk in older samples might be underestimated, and pastoralists are the more recent agriculturalist population (most used agriculture as a complementary diet), so it is only natural that selection had an impact over time in this aspect.
- Pastoralists have the smallest sample size (19 samples) and geographic range, so conclusions about this group are still less trustworthy.
- Genetic risk percentile ≠ Genomic health ≠ phenotypic health (not deterministic), and also disease-associated alleles in modern populations ≠ same effects in past environments.
To sum up, an interesting approach to studying genomic health with the scarce data available, but too many comparisons, with too many hypotheses being tested, which remind to a brute-force attack on data that can therefore yield statistically significant results anytime, anywhere.