58
A. Meyerdierks and F.O. Glöckner
of microorganisms in marine ecosystems (DeLong 2005, DeLong and Karl 2005,
Harrington et al. 2007, Sabehi et al. 2007, Yutin et al. 2007, Kagan et al. 2008).
When assembly fails, gene prediction has to cope with the additional problem of
fragmentation. When this is the case, BLASTx searches of the “environmental gene
tags” (EGT) (Tringe and Rubin 2005) against UniProt or Swiss-Prot can at least provide some information about the available functional space. Two strategies can be
used to compensate for these limitations: (1) pooling of sequence data from several
sampling sites to increase coverage, (2) establishment of a set of reference genomes
as templates for assembly and comparison (see the Gordon and Betty Moore Marine
Microbiology Project). Pooling has been used for the GOS datasets (Rusch et al.
2007), although this has incidentally caused a nightmare for ecologists. Building up
of the “oceans community genome” has led to the lost of data about specific adaptations to oceanic provinces and the correlation of these adaptations with habitat
parameters. This problem is compounded when short read technologies like pyrosequencing are used. A recent study on the influence of read length on functional
predictions showed that 100 bp fragments missed on average 72% of the BLASTx
hits found by long reads (∼750 bp) in the Sargasso, AMD (Tyson et al. 2004) and
Chesapeake bay datasets (Bench et al. 2007, Wommack et al. 2008). Nevertheless, it
has to be mentioned that a recent study by Mou et al. (Mou et al. 2008), which analysed populations involved in the metabolism of organic carbon compounds, showed
that valuable information can be obtained even with short read sequencing, provided
that targeted marine microcosm approaches are used.
2.4.7 Metagenome Descriptors for Comparative Metagenomics
Descriptors of phylogenetic and functional diversity can be used to obtain a better
understanding of the metagenome under investigation and to compare metagenomes
especially from environments with high biological diversity.
2.4.7.1 Phylogenetic Diversity
Several approaches can be used to describe phylogenetic diversity: (1) binning
and classification of the fragments as described above (2) phylogenetic analysis
of the ribosomal RNA genes (3) best BLAST hit mapping (4) analysis of singlecopy or equal-copy marker genes. Phylogenetic assignment of the ribosomal RNA
genes is rather straightforward. They can be mapped against the up to date ribosomal RNA (rRNA) sequence databases provided by SILVA (Pruesse et al. 2007)
or the Ribosomal Database Project II (Cole et al. 2007). The advantage of the
SILVA databases are that they provide quality checked and aligned sequences for
Eukarya as well. Furthermore, in addition to 16S/18S databases, 23S/28S rRNA
databases are provided for download at www.arb-silva.de. The SILVA compatible ARB software suite can be used for detailed phylogenetic tree reconstruction,
providing advanced alignment, tree reconstruction and visualization tools (Ludwig
et al. 2004).
A. Meyerdierks and F.O. Glöckner
of microorganisms in marine ecosystems (DeLong 2005, DeLong and Karl 2005,
Harrington et al. 2007, Sabehi et al. 2007, Yutin et al. 2007, Kagan et al. 2008).
When assembly fails, gene prediction has to cope with the additional problem of
fragmentation. When this is the case, BLASTx searches of the “environmental gene
tags” (EGT) (Tringe and Rubin 2005) against UniProt or Swiss-Prot can at least provide some information about the available functional space. Two strategies can be
used to compensate for these limitations: (1) pooling of sequence data from several
sampling sites to increase coverage, (2) establishment of a set of reference genomes
as templates for assembly and comparison (see the Gordon and Betty Moore Marine
Microbiology Project). Pooling has been used for the GOS datasets (Rusch et al.
2007), although this has incidentally caused a nightmare for ecologists. Building up
of the “oceans community genome” has led to the lost of data about specific adaptations to oceanic provinces and the correlation of these adaptations with habitat
parameters. This problem is compounded when short read technologies like pyrosequencing are used. A recent study on the influence of read length on functional
predictions showed that 100 bp fragments missed on average 72% of the BLASTx
hits found by long reads (∼750 bp) in the Sargasso, AMD (Tyson et al. 2004) and
Chesapeake bay datasets (Bench et al. 2007, Wommack et al. 2008). Nevertheless, it
has to be mentioned that a recent study by Mou et al. (Mou et al. 2008), which analysed populations involved in the metabolism of organic carbon compounds, showed
that valuable information can be obtained even with short read sequencing, provided
that targeted marine microcosm approaches are used.
2.4.7 Metagenome Descriptors for Comparative Metagenomics
Descriptors of phylogenetic and functional diversity can be used to obtain a better
understanding of the metagenome under investigation and to compare metagenomes
especially from environments with high biological diversity.
2.4.7.1 Phylogenetic Diversity
Several approaches can be used to describe phylogenetic diversity: (1) binning
and classification of the fragments as described above (2) phylogenetic analysis
of the ribosomal RNA genes (3) best BLAST hit mapping (4) analysis of singlecopy or equal-copy marker genes. Phylogenetic assignment of the ribosomal RNA
genes is rather straightforward. They can be mapped against the up to date ribosomal RNA (rRNA) sequence databases provided by SILVA (Pruesse et al. 2007)
or the Ribosomal Database Project II (Cole et al. 2007). The advantage of the
SILVA databases are that they provide quality checked and aligned sequences for
Eukarya as well. Furthermore, in addition to 16S/18S databases, 23S/28S rRNA
databases are provided for download at www.arb-silva.de. The SILVA compatible ARB software suite can be used for detailed phylogenetic tree reconstruction,
providing advanced alignment, tree reconstruction and visualization tools (Ludwig
et al. 2004).