This post describes the steps I took to assemble plastid genomes from low-coverage WGSS data. An overview of the approach can be found here.
Essentially, the method involves first mapping of quality-filtered reads to a reference plastid genome, and only selecting plastome-like reads from this mapping step for subsequent de-novo assembly. For the assembly step, I used the VELVET assembler, which performs well for small genomes and is quite fast.