Samples were collected from four oceanic regions (Figure 1). Briefly, the viral samples were concentrated on tangential flow filters (30–100-kD cutoff), distributed into 50-ml tubes and stored at 4 °C in the dark. A single sample was collected from the Sargasso Sea (labeled SAR) on 30 June 2005. Chloroform was added to this sample to stop microbial growth. Integrative samples, representing multiple sites and times, were assembled from the Gulf of Mexico (labeled GOM; 13 sites; 42 individual samples), the British Columbia coastal waters (labeled BBC; 38 sites; 85 individual samples), and the Arctic Ocean (labeled Arctic; 16 sites; 56 individual samples). These samples represent the combined viral assemblages of four oceanic regions over approximately one decade (sample details are described in Protocol S1).
Viral particles were purified using a combination of filtration and density-dependent centrifugation ([10 (link)]; http://scums.sdsu.edu/isolation.html, accessed 15 September 2006). The cesium chloride gradient was designed to recover virions with densities from 1.35 g ml−1 to 1.5 g ml−1. Viral DNA was isolated by a formamide/CTAB extraction [20 ], and the resulting DNA was amplified with Genomiphi and sequenced using pyrophosphate sequencing (454 Life Sciences, Branford, Connecticut, United States) [21 (link)] (see Protocol S1 for details on the technology). Each Genomiphi reaction started with 100–150 ng of DNA, above the 10 ng recommended by the manufacturer. A total of 181,044,179 base pairs (bp) of DNA sequence data was generated from the four libraries (SAR, 42 Mbp; GOM, 27 Mbp; BBC, 43 Mbp; and Arctic, 69 Mbp). The difference in library size was due to differences in number of successful reads during the pyrosequencing. The 1,768,297 sequences had an average length of 102 bp. The GOM, BBC, Arctic, and SAR metagenomes are deposited on the SDSU Center for Universal Microbe Sequencing website at (http://scums.sdsu.edu/phage/Oceans, accessed 15 September 2006).
Free full text: Click here