Figure 5: An example of complex apparent island configuration using an end-sequencing strategy.
Another approach is to sequence both ends of the random clones (typically
bp), and detect clone overlaps by comparing end sequences.
In other words, clones in an apparent island are linked via
their ends. The configuration of the apparent islands using this ad hoc strategy
is complicated, which makes it hard to analyze even for constant
clone length. (see Fig.5) In the fingerprinting scheme or
STS-content mapping, the apparent islands might overlap, but only to the extent
of length L. In other words, the necessary condition that a clone
starts an apparent island is simple, depending on only its neighboring
clones, and hence the intensity of the
island process can be computed by thinning argument. However, a clone
starts an apparent islands via the ends if and only if no clones
starting to its left overlap this clone at their ends, or
via the ends of other clones. Those ``indirect'' links depend on the
layout of the clones a lot, and this complicates the analysis.
Considering the first-order indirect links in the thinning argument
for constant length clones of the above setting, we can derive an upper bound for the expected
number of apparent islands (Yeh and Evans):
where E is the constant length of the end sequences, and
is the intensity of the clone process.
There are simulated results around ([16] or ask Ru-Fang), but no
successful theoretical analysis has yet been presented.