Primer extension permits no flexibility with respect to the location of the primers, primer length can be varied to adjust melting temperatures and the potential for formation of hairpins and dimers. Stable oligonucleotide secondary structures and primer dimers can affect signals. Moreover, structures involving primer 39 ends could result in the formation of templateindependent extension products. To minimize these effects, primer self-complementarity and dimerization potential were taken into account during the design process (see Materials and Methods; change in free energy dG values for the final primer set are listed in Table S1). This way the characteristics of the sequences surrounding the mutations impose limitations on primer size and product order (Table 2). Our multiplex assay relies on the simultaneous extension of several primers, subsets of which overlap and thus compete with each other. To examine the feasibility of the design, we first tested the primer set on DNA from normal individuals. The electropherograms were highly reproducible showing 15 extension product peaks corresponding to the normal HBB 69-25-0 web sequence and no unexpected peaks (Figure 1B). This indicates that all primers, including competing ones, produce detectable signals implying that the 15-plex primer set can be used for genotyping. Combined with the data on primer secondary structure and dimerization (Table S1), our results provide a comprehensive source of reference for the design of single-nucleotide extension primer mixes. We went on to test nine heterozygous samples, each carrying one of the mutations of interest (Figure 2). A specimen heterozygous for one of the interrogated mutations is expected to display two extra peaks (one for Codon 8 (-AA)) in addition to the 15 normal extension products. In most cases, the product from the mutant allele would 
migrate differently from the normal one, largely due to mass differences between dye-coupled nucleotides. Relative peak height can also vary significantly with the added nucleotide [32]. It is therefore important to confirm that all products, including these generated from mutant alleles, are detected and resolved by capillary electrophoresis. We observed that each mutation is manifested by well defined mutation-specific peaks in the electropherogram (Figure 2; Table S1). Normal genotype peaks are present but reduced in height, as expected for half the normal sequence dosage. These data show that the primer extension assay successfully detects the eight thalassemia mutations and the HbS hemoglobin variant. We next sought to assess the accuracy of the method by testing pre-genotyped samples, examining the proportion of correctly identified mutations (true positives) as well as the proportion of normal genotype calls obtained with non-carrier specimens (true negatives). We assayed a set of 128 reference chromosomes from normal individuals, mutation carriers and thalassemia major patients. Our results showed 100 agreement with the independently determined genotypes demonstrating that the new assay is highly 16574785 accurate (Table 3). Taken together, our analyses show that the multiplex assay is suitable for the detection of the nine Mediterranean mutations for diagnostic purposes.Genotyping Mediterranean HBB Gene MutationsTable 1. Panel of assayed HBB genetic variants.Mutation namea Codon 5 (-CT); CCT(Pro)-.CCodon 6 (-A); GAG(Glu)-.G-G beta 6(A3) Glu.Val Codon 8 (-AA); AAG(Lys)-.-G IVS-I-1 (G-.A) IVS-I-6 (T-.C) IVS-I-1.
