Whole genome sequencing of Turkish genomes reveals functional private alleles and impact of genetic interactions with Europe, Asia and Africa

Abstract

Background: Turkey is a crossroads of major population movements throughout history and has been a hotspot of cultural interactions. Several studies have investigated the complex population history of Turkey through a limited set of genetic markers. However, to date, there have been no studies to assess the genetic variation at the whole genome level using whole genome sequencing. Here, we present whole genome sequences of 16 Turkish individuals resequenced at high coverage (32×-48×).

Results: We show that the genetic variation of the contemporary Turkish population clusters with South European populations, as expected, but also shows signatures of relatively recent contribution from ancestral East Asian populations. In addition, we document a significant enrichment of non-synonymous private alleles, consistent with recent observations in European populations. A number of variants associated with skin color and total cholesterol levels show frequency differentiation between the Turkish populations and European populations. Furthermore, we have analyzed the 17q21.31 inversion polymorphism region (MAPT locus) and found increased allele frequency of 31.25% for H1/H2 inversion polymorphism when compared to European populations that show about 25% of allele frequency.

Conclusion: This study provides the first map of common genetic variation from 16 western Asian individuals and thus helps fill an important geographical gap in analyzing natural human variation and human migration. Our data will help develop population-specific experimental designs for studies investigating disease associations and demographic history in Turkey.

Figure 1

Brief summary of the Turkish Genome Project. (A) A map of Turkey showing provinces where volunteers were recruited in color. (B) Allele frequencies of the SNPs found in Turkish samples and annotated as novel vs. known (dbSNP135 + 1000 Genomes Project). (C) Functional annotation of the novel SNPs.

Figure 2

Population genetic relationships between Turkey and world-wide populations. (A) The first two principal components of the Turkish genome dataset combined with 16 individual population subsets from the 1000 Genomes Project dataset. The first and second components explain 6% and 5% of the total variance, respectively. (B) A population tree based on “Treemix” analysis. The populations included are as follows: Turkey (TUR); Toscani in Italia (TSI); Iberian populations in Spain (IBS); British from England and Scotland (GBR); Finnish from Finland (FIN); Utah residents with Northern and Western European ancestry (CEU); Han Chinese in Beijing, China (CHB); Japanese in Tokyo, Japan (JPT); Han Chinese South (CHS); Yoruba in Ibadan, Nigeria (YRI); Luhya in Webuye, Kenya (LWK). Populations with high degree of admixture (Native American and African American populations) were not included to simplify the analysis. The Yoruban population was used to root the tree. In total four migration events were estimated. The weights for the migration events predicted to originate from the East Asian branch into current-day Turkey was 0.217, from the ancestral Eurasian branch into the Turkey-Tuscan clade was 0.048, from the African branch into Iberia was 0.026, from the Japanese branch into Finland was 0.079. (C) The first three principal components of the Turkish genome dataset (left panels) and neighbor-joining trees of the 16 subjects (right panels). The upper and lower panels show the same data, except for being colored according to latitude and longitude of subject locations, respectively. The first, second and third principal components each explain ~7% of the total variance. Names of the provinces where each sample are recruited from are listed in Additional file 1: Table S1.

Figure 3

The proportion of synonymous and non-synonymous SNPs. SNPs that are found in a single chromosome among the 16 Turkish genomes (Rare) show a significant increase in the proportion of non-synonymous SNPs (NonSyn) to synonymous SNPs (Syn), as compared to all SNPs observed in the same population (p < 0.01, Chi-square with Yates’ correction). This increase is even more visible among SNPs that are seen in a single chromosome and are novel (i.e., not found in other databases) (NovelRare).

Figure 4

The allele frequency of GWAS SNPs. (A) The density distribution of allele frequency of GWAS SNPs among world populations. The distribution of GWAS SNPs in Turkey (TUR) is more similar to that of the European population (EUR) as compared to East Asian (ASN) and African (AFR) populations, indicating a greater proportion of ancestry sharing between TUR and EUR. Meanwhile, there is a higher proportion of common SNPs in both TUR and EUR than ASN and AFR, which is likely due to ascertainment bias in the GWAS studies as described in [35]. (B) The allele frequency of GWAS SNPs between Turkish (y-axis) and European (x-axis) populations. The red dots indicate SNPs in the 0.1 percentile (>0.345 allele frequency difference) of the absolute allele frequency distributions between Turkish and European populations. rs1129038 and rs12913832, both affecting HERC2 gene, are in strong linkage disequilibrium and, as such, have identical allele frequencies in European and Turkish populations. They are represented as two overlapping dots with 0.31 allele frequency in Turkish population (y-axis) and 0.71 allele frequencies in European population (x-axis). (C) The frequency distribution of derived and ancestral alleles for rs6712932 in Africa (ASW, LWK, YRI) Europe (CEU, FIN, GBR, IBS, TSI), Asia (CHB, CHS, JPT) and Americas (CLM, MXL, PUR) and Turkey (TUR). Note the increased frequency of the derived allele in the Turkish population.