Denaturing high-performance liquid chromatography (DHPLC) compares two or more chromosomes as a mixture of denatured and reannealed PCR amplicons, revealing the presence of a mutation by the differential retention of homo-and heteroduplex DNA on reversed-phase chromatography supports under partial denaturation. Temperature determines sensitivity, and its optimum can be predicted by computation. Single-nucleotide substitutions, deletions, and insertions have been detected successfully by on-line UV or fluorescence monitoring within 2-3 minutes in unpurified amplicons as large as 1.5 Kb. Sensitivity and specificity of DHPLC consistently exceed 96%. These features and its low cost make DHPLC one of the most powerful tools for the resequencing of the human and other genomes. Aside from its application to the mutational analysis of candidate genes, DHPLC has proven instrumental in elucidating human evolution and in the mapping of genes. Employing completely denaturing conditions, the utility of DHPLC has been extended to the genotyping of known polymorphisms by utilizing the ability of poly
INTRODUCTIONThe analysis of DNA sequence variation is of fundamental importance in genetic studies [Lander, 1996]. With the increasing availability of primary sequence from genomes of human and other organisms, the implementation of sensitive, efficient, and inexpensive technologies for detecting variation that can match the pace of primary sequencing becomes critical. The least expensive method is in silico mining of publicly available databases that contain large amounts of overlapping sequence data from heterogeneous samples. The specificity of this approach which ranges from 55-82% [Buetow et al., 1999;Picoult-Newberg et al., 1999;Cox et al., 2000] is sufficient for the rapid assembly of SNPs for whole-genome linkage disequilibrium studies. But its low sensitivity of 27% [Cox et al., 2000] is insufficient for performing association studies 440 XIAO AND OEFNER on candidate genes. In addition, existing databases are heavily biased towards Caucasian-specific SNPs [Cox et al., 2001].Technologies suitable for the experimental discovery of SNPs and disease causing mutations should be capable of fully automated highthroughput analysis that ideally does not require modified PCR primers, customized specific reagent arrays, detection labels, or any sample pretreatment other than PCR. At present, numerous techniques for DNA mutational analysis are available [Cotton, 1997]. None of these methods meets this challenge. They are either formatted for manual use only and are technically challenging (e.g., denaturing gradient gel electrophoresis (DGGE), chemical cleavage) or limited by the sensitivity of detection (e.g., conformation sensitive gradient gel electrophoresis (CSGE)) [White et al., 1992;Vidal-Puig and Moller, 1994;Couch and Weber, 1996]. Other approaches such as immobilized DNA hybridization arrays and enzymatic mismatch cleavage assays, while promising, still have significant false positive signal, as well as high cost per assay ...