TKGWV2: an ancient DNA relatedness pipeline for ultra-low coverage whole genome shotgun data

Abstract

Estimation of genetically related individuals is playing an increasingly important role in the ancient DNA field. In recent years, the numbers of sequenced individuals from single sites have been increasing, reflecting a growing interest in understanding the familial and social organisation of ancient populations. Although a few different methods have been specifically developed for ancient DNA, namely to tackle issues such as low-coverage homozygous data, they require a 0.1–1× minimum average genomic coverage per analysed pair of individuals. Here we present an updated version of a method that enables estimates of 1st and 2nd-degrees of relatedness with as little as 0.026× average coverage, or around 18,000 SNPs from 1.3 million aligned reads per sample with average length of 62 bp—four times less data than 0.1× coverage at similar read lengths. By using simulated data to estimate false positive error rates, we further show that a threshold even as low as 0.012×, or around 4000 SNPs from 600,000 reads, will always show 1st-degree relationships as related. Lastly, by applying this method to published data, we are able to identify previously undocumented relationships using individuals that had been excluded from prior kinship analysis due to their very low coverage. This methodological improvement has the potential to enable relatedness estimation on ancient whole genome shotgun data during routine low-coverage screening, and therefore improve project management when decisions need to be made on which individuals are to be further sequenced.

Genome-wide analysis of a collective grave from Mentesh Tepe provides insight into the population structure of early neolithic population in the South Caucasus

“…8 and Supplementary Data 5 ), and with a high number of overlapping SNPs ( n = 17864) which is well over READ’s satisfactory threshold of around 1500 SNPs. These results were subsequently confirmed using TKGWV2 39 , an alternative kinship estimation method which provided us with the same findings (Supplementary Fig. 9 ).…”

“…9 ). As per recommendations from the authors, the final BAM files from all Mentesh individuals were downsampled to around 1.8M reads beforehand 39 . We then performed the analysis using the 22M nonfixed biallelic variant set provided by Fernandes et al 39 .…”

“…As per recommendations from the authors, the final BAM files from all Mentesh individuals were downsampled to around 1.8M reads beforehand 39 . We then performed the analysis using the 22M nonfixed biallelic variant set provided by Fernandes et al 39 . To assess confidence in the raw relatedness coefficients given by the method, we obtained posterior probabilities by simulating distribution curves for each tested pair of individuals ( n = 6000 simulation replicates).…”

Neolithic genomic data from southern France showcase intensified interactions with hunter-gatherer communities

“…To complement biological relatedness analysis, we used the TKGWV2 pipeline that can be used to infer pairwise relatedness for individuals presenting very low coverage ( Fernandes et al., 2021 ). We used default parameters, i.e .…”

READv2: Advanced and user-friendly detection of biological relatedness in archaeogenomics

“…Recent years have seen the introduction or application of more advanced methods into the field which work with lower amounts of data (Table S1), provide resolution for more distant degrees of relatedness, and/or are able to differentiate between different types of relationships for the same degree (e.g. parent-offspring versus full siblings) [29][30][31][32][33][34][35].…”