Elucidation of the molecular bases of Mendelian disease has provided a rich resource for understanding genetic mechanisms, protein functions, the behavior of biological systems and mechanisms of disease 
. Despite intense efforts with a variety of approaches, however, human geneticists have so far identified only ~2,400 genes responsible for Mendelian phenotypes, or about 11% of the total number of protein coding genes in our genome. Currently, OMIM, a catalog of Mendelian disorders 
, lists >1,500 mapped Mendelian disorders for which the gene has yet to be identified, and practicing clinical geneticists know that there is an untold number of families with Mendelian disorders for which a molecular explanation or even clear mapping information has yet to be accomplished. Challenges that prevent harvesting this trove of biomedical information include the rarity of each disorder, small family sizes, reduced reproductive fitness of affected individuals, locus heterogeneity and diagnostic tools that query only a fraction of all biological systems 
The recent development of massively parallel DNA sequencing technologies has reduced the cost and increased the throughput of large-scale sequencing (LSS) and provides a new and potentially powerful way to identify virtually all of the mutations responsible for Mendelian disorders 
. Indeed, at least two groups have used LSS coupled with hybridization strategies to “capture” the majority of known exons (the “exome”) for protein coding genes to identify genes responsible for three Mendelian disorders 
. While there are many good reasons to use whole-exome sequencing (WES), including the lower cost (currently ~5-fold) and the fact that exon variation is the most readily interpreted, it is also clear that WES will miss mutations of interest, including those variants that are either in exons that are not captured, are in non-exonic regulatory regions, or are structural variants. At least 1.4% of the disease variants listed in the Human Gene Mutation Database <http://www.hgmd.cf.ac.uk/ac/index.php
> are in regulatory sequences and this is likely to be an underestimate given the traditional strategies used for mutation detection (largely PCR amplification and sequencing of exons). The same database lists about 7.5% of disease variants as structural variants. WES also requires higher average coverage levels than WGS, both because of the variable success of capturing different regions and because of “allelic imbalance” where one allele is preferentially captured over the other. For these reasons we elected to employ a WGS strategy.
Both WES and WGS identify a large number of sequence variants when compared to the reference sequence, making it important to prioritize variants. This can be based upon both the assessment of the likelihood of the variants being functional, especially in the WGS setting (Shianna et al. submitted), and on their frequency in healthy control populations. These approaches were used in the WES study that discovered the gene for Miller syndrome (OMIM 263750) by identifying functional mutations in the same gene in each of four unrelated patients and no controls (6). In addition to these approaches, it also is possible to utilize classical genetic strategies that depend on family structure and inheritance patterns to prioritize certain genomic regions. In our case, we show that it is possible to combine partial linkage information with other criteria for prioritizing variants to identify the genetic basis of a rare autosomal dominant disorder (metachondromatosis, OMIM 156250).
The condition we have studied, metachondromatosis (MC, OMIM 156250), is an autosomal dominant condition characterized by exostoses, commonly of the bones of the hands and feet, and enchondromas of the metaphyses of long bones and iliac crest. It was first described by Maroteaux in 1971, based on clinical observation of the six affected individuals from two families 
. Shortly thereafter, Lachman reported a young male with enlarging, painless, hard lumps on multiple fingers with concurrent long bone metaphyseal and iliac irregularities, both proven by histopathology to be classic exostoses and enchondroma, respectively 
. The enchondromas often have a “striated” appearance in radiographs, and in MC both types of lesions typically appear in childhood and may regress or even resolve over several years 
(). This phenomenon likely contributes to the incomplete penetrance that has been described in MC families 
. The exostoses of MC differ in location, orientation and duration from those observed in a related set of phenotypes known as the hereditary multiple exostoses syndromes (MES I and II, OMIM 133700 and 133701). The exostoses of MESs rarely resolve and can cause permanent deformity 
. The osteochondromas of MESs typically point away from the adjacent epiphysis and rarely affect the hands or feet, while those of MC point toward the epiphyses and usually present on the hands and feet 
. Though palpable, the exostoses of MC may not be calcified and therefore may be radiolucent 
, in part depending on the timing of the clinical exam and radiography in the lifespan of a given lesion. The enchondromas of MC are similar to those of Ollier disease (OMIM 166000, also known as multiple enchondromatosis) but the latter disorder usually lacks exostoses. Mutations in EXT1
, located respectively at 8q24 and 11p11–p12, have been identified in 70% of MES cases 
Manifestations of metachondromatosis.