Mapping undetected mutations within a gene-evidence for two preferential regions in the DMD gene

A maximum-likelihood method is developed to estimate the frequency distribution of undetected mutations (presumably point mutations, small deletions, insertions) along a gene, where the gene extends over a long stretch of DNA. In each family, the point of the mutation is potentially at a different location within the gene. In this sense, there is genetic heterogeneity among families and the method estimates the proportion of families whose mutation is at (or in the vicinity of) a given point inside the gene. Our method is applied to a sample of 75 families with Duchenne muscular dystrophy in which the disease mutation remained undetected. We find two preferential regions for these undetected mutations, with an estimated 85% of families having their mutation in one region and the remaining 15% of families in the other. The new method is expected to be useful in finding small mutations in any of the currently known large genes.