Once extracted, ancient DNA is fragmented — it comes in short pieces rather than the long stretches found in living cells. The fragments are also chemically damaged at their ends, bearing the characteristic signature of deamination (the conversion of cytosine to uracil) that distinguishes ancient DNA from modern contamination. These damage patterns are diagnostic: a sample with the expected ancient DNA damage pattern is more likely to represent genuine ancient material than a sample without such patterns.

Next-generation sequencing technology processes millions of these short, damaged fragments simultaneously, generating enormous quantities of sequence data. Computational methods then sort the genuine ancient human sequences from the contaminating modern sequences, using the damage patterns and other characteristics to discriminate between them.

For kinship analysis, the key measurement is the extent of identical-by-descent (IBD) sharing between pairs of individuals. When two people share a recent common ancestor, they inherit from that ancestor identical segments of DNA — the same sequence, in the same location in the genome. The more recently they shared an ancestor, the more of these shared segments they will have, and the longer those shared segments will be (because recombination has had less time to break them up). By measuring the total length and number of shared segments between two individuals, geneticists can estimate whether they are parent-child (the most closely related, sharing approximately half the genome), full siblings (also approximately half), half-siblings (approximately a quarter), and so on through increasingly distant relationships.

Applied to a set of thirty-five individuals, this method requires comparing every possible pair — a total of several hundred pairwise comparisons. For each pair, the researchers estimate the degree of relationship, then try to construct a consistent pedigree that accounts for all the relationships simultaneously. The pedigree must be consistent: if person A is the parent of person B, and person B is the parent of person C, the genetic data should reflect both of those relationships and also the implied grandparent relationship between A and C.

ADVERTISEMENT
Advertisement
Share.