Genomes and metagenomes from marine environments were selected for import from the NCBI databases 9 into a local PostgreSQL/PostGIS database 10, according to the following criteria: i) the DNA sequence must be of marine bacterial or archaeal origin; ii) sequence quality must be high (i.e. sequencing coverage of at least eight fold); iii) marker and single genes are rejected; and iv) the geographic origin of the DNA sequences must be known precisely (e.g. from the original publication). Lower quality sequences (draft genomes and short metagenomics reads) will be included in future releases.

Geographic locations were stored in our database for accepted DNA samples. Moreover, on-site contextual (meta-)data, such as physical and chemical parameters at the sampling site, were retrieved manually from the original publications and additional web pages when available. This manual curation step is crucial in order to reliably link on-site contextual data to DNA sequences. Moreover, having the exact geographic position for each sample in our database allows the interpolation of environmental parameters from worldwide data sets. Currently, the following global oceanic physical and chemical parameters are integrated into our database from the WOA data set (World Ocean Atlas): temperature, nitrate, phosphate, oxygen and silicate concentration, as well as salinity 11.

The MetaLook interface is a locally running client based on the Java 3D API 12, started using the Java Web Start technology from the megx.net data portal 7. The starting point of the interface is a 3D workbench displaying a world map with the sampling sites of genomic and metagenomic studies available in our database (Fig. 1). The 3D approach allows displaying larger amounts of data and interconnections than a classical 2D visualisation 13. Within MetaLook, the user can sort the corresponding DNA sequence data into so-called environmental containers, which are flexible entities grouping data according to specific criteria, such as habitat types, ocean water depth or physical and chemical parameters. This custom data classification allows the user to define ecological niches according to specific biological questions (Fig. 2). The DNA sequence fragments grouped into the containers can be visualised on the workbench for browsing and comparing (Fig. 3). Moreover, each container can be searched for specific genes based on their annotations. Search results are shown graphically in their genomic context. The DNA or protein sequence of each gene can be displayed or easily downloaded from the database. All DNA sequences in a container can be downloaded in batch mode. Custom sequences can be imported into the MetaLook interface in FASTA format.

Any protein encoding gene from our georeferenced database or user-imported sequences can be used as a query for a BLASTP run 14 against the genes grouped into user-defined environmental containers. The BLASTP analysis is started from the MetaLook interface (client) and runs on the centralised server. The results are shown graphically using 3D connectors between the query gene and the containers with sequence matches (Fig. 4). This representation reveals the distribution of similar genes in the user-defined habitats. The results are saved in a result panel for detailed investigation, showing the habitat parameters of each match, the corresponding BLASTP e-value, and sequence alignment (Fig. 5a, b).

Some interesting DNA sequence tools making use of 3D are currently available. Sockeye is a 3D environment for comparative genomics allowing simultaneous visualisation of the annotations of different eukaryotic organisms 15. The Correlogo server is a tool to display DNA sequence alignments using 3D sequence logos 16. The Walrus graph visualisation tool allows visualisation of very large phylogenetic trees in a hyperbolic space 17. These examples show the benefits of advanced visualisation tools for DNA research and the management of large data sets. However, within this context, MetaLook is unique in its orientation toward environmental genomics, geographic and contextual data integration.

In microorganisms, methanogenesis is a form of microbial anaerobic respiration leading to the formation of methane. Recent experimental and genomic data support the hypothesis that anaerobic oxidation of methane (AOM) is using a reverse-methanogenesis pathway 181920. Such biochemical processes are crucial in the environment, as methane is an important greenhouse gas contributing to global warming. One of the key genes of methanogenesis and AOM is mcr, encoding a methyl-coenzyme-M reductase (Mcr). The distribution of mcr in the environment was visualised by MetaLook with the following steps: i) predefined environmental containers were created from the world map, grouping sediment and water samples by depth (Fig. 2); ii) a text search for the gene "mcr" was performed; iii) Mcr protein sequences (e.g. McrB, [Genbank: AAB98847]) were blasted against all containers (BLASTP, e-value cut-off 10^-10). The results show that within the georeferenced marine bacteria and archaea currently available in our database, genes encoding Mcr are only found in sediments. Although expected, this observation shows that mcr genes are habitat specific, which is consistent with the strictly anaerobic nature of methanogenesis and the AOM process.

Another key gene of the methanogenesis and AOM processes is mch, encoding a methenyl-tetrahydromethanopterin cyclohydrolase (Mch). Interestingly, this gene was reported in some proteobacteria and planctomycetes, where an archaea-like C1 metabolism appears to be present 21. Following the same procedure used for mcr (see above) revealed, as expected, that the mch gene is not only present in genomes and metagenomes originating from sediments, but is also found in the genome of at least one sea water column bacterium, the planctomycete Rhodopirellula baltica SH 1^T from the Baltic Sea 22 (e.g. [GenBank: CAD74990]). Furthermore, this analysis showed that mch is also found in the high-throughput metagenomics data set of the Sargasso Sea 23, suggesting an even more widespread distribution of this gene in the environment. Hence, the analysis of the habitat specificity of mcr and mch revealed differential environmental distribution of genes relevant for major biochemical processes involved in the global cycling of carbon.

Solar UV-light induces pyrimidine dimers in genomic material, leading to enhanced mutation rates. Photolyases are proteins involved in a light-dependant, single-step DNA repair mechanisms, which protect microorganisms against this destructive effect 24. Comparative analysis of the genomes of three Prochlorococcus marinus strains, one of the most abundant phototrophic prokaryote in the ocean, previously reported the presence of photolyase encoding genes (phr) in the high-light ecotype, and its absence in the low-light ecotypes (water depth: 5 m and 120/135 m, respectively) 25. This finding suggests that for this particular species, the phr gene is lost if an organism is exposed to little or no UV-light. As no DNA pyrimidine dimers should form where no UV-light stress occurs, the phr gene is not expected in the deep layers of the ocean.

To systematically test the occurrence of the phr gene in the marine environment, a phr gene with experimental evidence (Escherichia coli K-12, [Swiss-Prot: P00914]) was imported into the MetaLook interface and searched against predefined environmental containers with the BLASTP algorithm (e-value cut-off 10^-10). Some sequence hits in the top layers of the ocean were found, as expected (e.g. Prochlorococcus marinus MED4, [Genbank: CAE18744] and Rhodopirellula baltica SH 1^T, [Genbank: CAD77347]). Moreover, unexpected sequences from deep-sea water (hot vent) and coastal sediments were also hit by this analysis (Idiomarina loihiensis L2TR, [GenBank: AAV82228] and Hahella chejuensis KCTC 2396, [GenBank: ABC28582]) 2627 (Fig. 3). These genes are likely to be functional, with full-length BLASTP alignments and excellent statistical support, with e-values below 10^-100 (Fig. 4a, b). Such unexpected occurrence of genes encoding photolyases in these environments might be explained by: i) the presence of allochthonous organisms 28, ii) residual phr genes awaiting deletion in organisms recently adapted to deep-sea or sediment environments, or iii) the possible need for protective mechanisms against geothermal light, even if the dominant wavelengths are not in the UV range 29.

The availability of worldwide physical and chemical parameters linked to DNA sequences opens the way to multivariate analysis. This approach will be crucial as more georeferenced genomic and metagenomic samples become available. The integration of low quality sequences (e.g. single reads from metagenomics) and biodiversity markers (e.g. ribosomal RNA genes) in our geographic-centric system is also a follow-up perspective.