Jun Yu, Gane Ka-Shu Wong, Edward C. Thayer, and Maynard Olson
Department of Molecular Biotechnology, University of Washington, Seattle WA 98195
We have reduced to practice a standardized method for generating unusually accurate, high-resolution restriction maps supported by densely overlapping cloned coverage of the target region. In the present implementation, the starting point is a yeast-artificial chromosome (YAC) contig that has been mapped by STS-content mapping. A subset of the YACs is subcloned at high redundancy into cosmids. The method is equally compatible with other high-level mapping strategies and cloning systems. The main prerequisite is a source of densely overlapping clones from the target region whose DNA can be characterized by cleavage with a variety of restriction enzymes that produce a moderate number (i.e., approx. 5-15) of fragments.
Two principal software systems underlie multiple-complete-digest (MCD) mapping. Will Gillett (Department of Computer Science, Washington University) has developed a sophisticated map-assembly tool, which converts fragment-size lists for a set of densely overlapping clones, each digested independently with an arbitrary number of enzymes, into MCD maps[1]. To provide input for the Gillett map assembler, we have developed a fully automated image-analysis tool that takes digitized images of stained agarose gels on which restriction digests have been fractionated and produces fragment-size lists. Without interactive intervention, this cross-platform tool presently interprets >95% of gel lanes correctly (i.e., no false-positive or false-negative fragment calls) with a typical fragment-sizing accuracy of +1-2%.
We have reduced this system to practice in a collaboration with Daniel Geraghty (Fred Hutchinson Cancer Research Center), starting with a YAC contig that he has produced for the class I region of the HLA locus[2]. A subset of the YACs is subcloned into cosmids at 20-fold redundancy. Cosmids containing human DNA are selected by colony hybridization using labeled human DNA as the probe. Each cosmid is independently digested with 3 different restriction enzymes that have 6-bp recognition sites. The restriction fragments are separated by size on an agarose gel, which is post-stained (SYBR Green from Molecular Probes), imaged with a fluorescence scanner (Molecular Dynamics FluorImager 575), and analyzed with the software package described above. Vector-insert fusion fragments are identified by gel-transfer hybridization. All insert fragments larger than 500 bp are used by the Gillett mapping software to construct maps which position both restriction sites and clone ends. Initial mapping experience now includes several HLA YACs. In some cases, continuous sequence data are now available from D. Geraghty for mapped regions longer than 100 kbp. In these cases, there is near-perfect agreement between the MCD- and sequence-derived maps. Hence, MCD mapping provides a powerful check on shotgun-sequence assemblies, and allows precise localization of any residual gaps in assembled sequence.
*Supported by a grant from the Director, Office of Energy Research, Office of Health and Environmental Research of the U.S. Department of Energy under contract DE-FG06-92ER61487.
[1] W. Gillett, L. Hanks, G. K.-S. Wong, J. Yu, R. Lim, and M.V. Olson, Genomics, in press 1996).
[2] D. Geraghty, Curr. Opin. Immunol. 5, 3-7 (1993).