Reviews of highthroughput sequencing technologies and assembly tools can be found elsewhere. In addition to short read assemblers, there are specialized tools for assembling longer pyrosequencing reads, such as CABOG. Even the current assemblies of important model organisms are subject to continuing finishing processes; for example, recent improvements in the mouse genome assembly added 267 Mb of previously missing or misassembled sequence. Efforts to finish shotgun-based vertebrate genome assemblies are further complicated by a high amount of species-specific variability regarding mis-assembly and gap characteristics, making it challenging to apply standardized finishing strategies. Some promising approaches for tackling the problem of high-throughput sequence assembly by using a closely related reference genome have been proposed, including gene-boosted assembly and assisted assembly. Further complicating the picture are the error profiles of the various new sequencing technologies and associated platforms. These profiles have not been adequately characterized in the literature, and they appear to be changing with every iteration of a given platform. To the best of our knowledge, there has been only anecdotal evidence on the impact of the resulting error rates on the available assembly tools. The work described below has been motivated by our participation in the USDA/MARS/IBM consortium whose goal is to sequence and analyze the genome of Theobroma cacao with an estimated length of approximately 400 M bases. One of the questions that arose in the context of the project is whether the capabilities of today’s high-throughput sequencing platforms are such that a de novo assembly of T. cacao from short reads is feasible. Short read lengths present formidable challenges for de novo genome assembly because several valid alignments can exist for a given set of very short sequences. In principle, one of those possibilities corresponds to the target genome sequence. The number of alignment possibilities depends on the length of overlap that is required to align the ends of two sequences. There are also limits to the quality of the assembly results that can be achieved: it is not possible to determine the exact size of tandem repeats that are longer than the read length. Also distinguishing between two near-exact copies of the same repeat in different parts of the genome may not be possible, since short reads do not necessarily provide enough sequence context to determine the relative position of the read in the genome. Adding information from paired reads with large insert sizes can potentially assist in determining the correct origin of repeat copies and can also help in scaffolding contigs into longer stretches of ordered sequence. Highly fragmented assemblies with repeat expansions and collapses, and falsely joined sequences can be characteristic of short read assembly results on repeat-rich genomes. Wortmannin clinical trial Clearly, these complications continue to persist even in the presence of high sequencing coverage. As outlined above, there are several challenges and sources of error associated with genome.