RESUMO
Recombination patterns can be indirectly inferred by means of linkage disequilibrium (LD) estimates, since LD is negatively correlated with genetic distance. However, LD does not necessarily have absolute correspondence with genetic distance. We estimated LD at 5 loci located in the 21q22.3 region. These STRs (D21S1440, D21S168, D21S1260, D21S1446, and D21S1411) covered 8.81 Mb of the 21q22.3 region. They were genotyped by conventional PCR. Similar size samples previously validated by sequencing were used as a genotyping control. Three hundred and sixty-nine individuals (62 families) living in Guadalajara, Mexico, were included. As an inclusion criterion, each family had a positive paternity test by autosomal markers for the CODIS core loci. Two hundred and thirty phase known haplotypes were identified by familial segregation. Only those haplotypes whose frequency was higher than 4% were taken into account for LD estimation, expressed as Lewontin's D' coefficient and Bonferroni's correction P values. For all 5 loci, the genetic distributions were in agreement with Hardy-Weinberg expectations. Heterozygosity and haplotype diversity were ≥ 0.69 and 99.58%, respectively. D21S1440-D21S168 (4.51 cM) and D21S1446-D21S1411 (4.58 cM) marker haplotype frequencies were significantly different from those expected by random distribution. The remaining haplotypes, including those with minimal inter-distance (D21S1260-D21S1446, 1.44 Mb), did not show LD. The 5 STRs at the 21q22.3 region in this Mexican population showed a non-homogeneous LD pattern, which demonstrates that recombination or linkage should not be assumed solely on the basis of genetic distance.