This approach is not dependent upon existing BAC contig information, and so should be particularly useful in the application to previously uncharacterized genomes. Nevertheless, the method may be used to independently validate a BAC contig map or a minimal tiling path obtained by intensive genomic sequence determination.

We provide a detailed probabilistic analysis to characterize the outcome of a single hybridization experiment and what information can be garnered about the physical distance between any pair of probes. This analysis then leads to a formulation of a likelihood optimization problem whose solution leads to the relative probe locations. After reformulating the optimization problem in a graphtheoretic setting and by exploiting the underlying probabilistic structure, we develop an efficient approximation algorithm for our original problem. We have implemented the algorithm and conducted several experiments for varied sets of parameters. Our empirical results are highly promising and are reported here as well. We also explore how the probabilistic analysis and algorithmic efficiency issues affect the design of the underlying biochemical experiments.