What can you do with 0.1× genome coverage? A case study based on a genome survey of the scuttle fly Megaselia scalaris (Phoridae)

Random pyrosequencing of the M. scalaris genome provided a depth of coverage (0.05x-0.1x) much lower than typical GSS studies. We demonstrate that, even with extremely low-coverage sequencing, bioinformatics approaches can yield extensive information about functional and repetitive elements. We also use our GSS data to develop genomic resources such as a nearly complete mitochondrial genome sequence and microsatellite markers for M. scalaris.We conclude that low-coverage genome surveys are effective at generating useful information about organisms currently lacking genomic sequence data.Next-generation sequencing technologies, such as the 454 (Roche Applied Science) and Solexa (Illumina) platforms, provide researchers working on emerging- and non-model species an affordable means of addressing a wide range of questions [1]. While completely assembled genomes of non-model organisms may not be easily obtained, very low-coverage shotgun sequencing can be used for various applications, such as to identify microsatellites for population genetic analyses [2,3]. Low coverage genome survey sequences (GSS) can also provide information about gene content, polymorphisms, functional elements and repetitive elements [4].In support of the utility of low-coverage sequencing, simulations have shown that most of the coding sequence in a genome can be surveyed with less than 2× genomic coverage [5]. As a case in point, the original 1.5× assembly of the dog genome provided partial sequence of thousands of dog orthologs of human genes [6]. At even lower coverage, Wernersson et al. [7] were able to recover 38% of the coding fraction of the mouse-human alignment with only 0.66× coverage of the pig genome. Furthermore, low-coverage survey sequencing appears to be an efficient way of identifying common repetitive DNA sequences. For example, a large fraction of the repetitive DNA sequences in the complex, highly repetitive barley genome of were computationally identified with only ~10% of t