Nucleic acid sequences that differ in abundance between two population
s (target sequences) can be cloned by multiple rounds of subtractive h
ybridization and amplification by PCR. These sequences can be cDNAs re
presenting up-regulated mRNAs, or genomic DNAs from deletion mutants.
We have derived an equation that describes the recovery of such sequen
ces, and have used this to simulate the outcome of up to 10 rounds of
subtractive hybridization and PCR amplification. When the model was te
sted by comparing its predictions with the published results from geno
mic and cDNA subtractions, the predictions of the model were generally
in good agreement with the published data. We have modelled the outco
mes of genomic subtractions, for a variety of genomes, and have used i
t to compare Various strategies for enriching targets. The model predi
cts that for genomes of less than 5 x 10(8) bp, deletions of as small
as 1 kbp should represent >99% of the DNA after three to six rounds of
hybridization (depending on the enrichment procedure). As genomes inc
rease in size, the kinetics of hybridization become an important limit
ing factor. However, even for genomes as large as 3 x 10(9) bp, it sho
uld be possible to isolate deletions of 5 kbp using the appropriate co
nditions. These simulations suggest that such methods offer a realisti
c alternative to chromosome walking for identifying genomic deletions
for which there are known phenotypes, thereby considerably reducing ti
me and effort. For cDNA subtractive hybridization, the model predicts
that after six rounds of hybridization, sequences that do not differ i
n abundance between the tester and driver populations (the background)
will represent <1% of the subtracted population, and even quite modes
tly upregulated cDNAs should be successfully enriched. Where several u
p-regulated cDNAs are present, the predicted final representation is d
ependent on both the initial abundance and the degree of up-regulation
.