We have developed a new procedure that relies on an array of cross-hybridization tests to order a set of random clones into a contig. The method, called cross-screening, uses each clone as a target and its end sequences as probes, in a matrix of reciprocal cross-hybridization tests performed on a single blot. The relationships among the clones are determined rapidly from the pairwise tests, allowing clone order to be determined directly. We have applied this technique to DNAs from a set of overlapping ~ clones from Drosophila chromosome 4. The location and orientation of each clone derived from the cross-screening data was that expected from the map assembled from overlapping restriction sites and chromosomal walking. The procedure provided additional information on a previously unknown, internally repeated DNA sequence. To demonstrate the general utility of the procedure, we have applied it to a previously described clone set within a contig in region 22q12 of human chromosome 22. The correct relative position and orientation of these clones were derived from the cross-screening data without knowledge of, or reference to, any nucleotide sequence or restriction site analysis of the DNA concerned. The cross-screening procedure is fast, economical, and robust and allows clone overlaps to be determined efficiently, with minimal interference from repeated DNA sequences. This new procedure is specifically designed for small groups of overlapping clones [tens to hundreds] and should facilitate the ordering of subclone libraries derived from small chromosomes or the large cloned inserts carried in YAC, BAC, and PI vectors.

Chromosome 4 from Drosophila melanogaster has several unusual features that distinguish it from the other chromosomes. These include a diffuse appearance in salivary gland polytene chromosomes, an absence of recombination, and the variegated expression of P-element transgenes. As part of a larger project to understand these properties, we are assembling a physical map of this chromosome. Here we report the sequence of two cosmids representing ∼5% of the polytenized region. Both cosmid clones contain numerous repeated DNA sequences, as identified by cross hybridization with labeled genomic DNA, BLAST searches, and dot matrix analysis, which are positioned between and within the transcribed sequences. The repetitive sequences include three copies of the mobile element Hoppel, one copy of the mobile element HB, and 18 DINE repeats. DINE is a novel, short repeated sequence dispersed throughout both cosmid sequences. One cosmid includes the previously described cubitus interruptus(ci) gene and two new genes: that a gene with a predicted amino acid sequence similar to ribosomal protein S3a which is consistent with the Minute(4)101 locus thought to be in the region, and a novel member of the protein family that includes plexin and met–hepatocyte growth factor receptor. The other cosmid contains only the two short 5′-most exons from thezinc-finger-homolog-2 (zfh-2) gene. This is the first extensive sequence analysis of noncoding DNA from chromosome 4. The distribution of the various repeats suggests its organization is similar to the β-heterochromatic regions near the base of the major chromosome arms. Such a pattern may account for the diffuse banding of the polytene chromosome 4 and the variegation of many P-element transgenes on the chromosome.