SILVA, RDP, Greengenes, NCBI and OTT - how do these taxonomies compare?
Balvočiūtė M, Huson DH. · 2017
Read SILVA, RDP, Greengenes, NCBI and OTT - how do these taxonomies compare? by Balvočiūtė M, Huson DH. (2017) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc5374703.txt.
Introduction
Monika Balvočiūtė and Daniel H. Huson compared the family trees used to name microbes in BMC Genomics in 2017. A core step in microbe reading is sorting reads into named kinds. Four large name systems rule that step, each built in its own way, plus a fifth that maps the whole tree of life. No one had charted how well the five line up, or how to move results between them.
The pair built a method and a program that carry a named kind from one system onto another by name match. Three of the systems, SILVA, RDP, and Greengenes, fit neatly into the NCBI system, and all four fit into the open tree of life. The hard way ran in reverse: squeezing the two large systems into the smaller ones lost or clashed on many names. Their maps let labs merge work done under unlike systems.
Catalog record
| CREATOR | Balvočiūtė M, Huson DH. |
|---|---|
| YEAR | 2017 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
SILVA, RDP, Greengenes, NCBI and OTT — how do these taxonomies compare? By Monika Balvočiūtė, Daniel H. Huson Published in: BMC Genomics DOI: 10.1186/s12864-017-3501-4 ABSTRACT Background A key step in microbiome sequencing analysis is read assignment to taxonomic units. This is often performed using one of four taxonomic classifications, namely SILVA, RDP, Greengenes or NCBI. It is unclear how similar these are and how to compare analysis results that are based on different taxonomies. Results We provide a method and software for mapping taxonomic entities from one taxonomy onto another. We use it to compare the four taxonomies and the Open Tree of life Taxonomy (OTT). Conclusions While we find that SILVA, RDP and Greengenes map well into NCBI, and all four map well into the OTT, mapping the two larger taxonomies on to the smaller ones is problematic. Electronic supplementary material The online version of this article (doi:10.1186/s12864-017-3501-4) contains supplementary material, which is available to authorized users. BACKGROUND Microbiome sequencing analysis is concerned with sequencing DNA from microorganisms living in certain environments without cultivating them in laboratory. In a typical taxonomy guided approach [1], sequencing reads are first binned into taxonomic units and then the microbial composition of samples is analyzed and compared in detail (see Fig. 1). Fig. 1Basic taxonomic binning workflow The two main technical ingredients of taxonomic analysis are the reference taxonomy used and the binning approach employed. Binning is usually performed either by aligning reads against reference sequences (e.g. [2]) or using k-mer based techniques (e.g. [3]). Taxonomic binning of 16S reads is usually based on one of these four taxonomies: SILVA [4], RDP [5], Greengenes [6] or NCBI [7]. How important is the choice of reference taxonomy, given the known inconsistencies of microbial classifications [8]? To address this, the aim of this paper is to determine how similar these four taxonomies are, and whether results obtained using one classification can easily be carried over to another. We define and explore an algorithm for mapping one taxonomy into another. This method allows us to compare taxonomies and is the basis for a tool that makes analyses on different classifications comparable to each other by mapping them onto a common taxonomy. While our main focus is on the four most popular taxonomic trees, we also consider the recently published Open Tree of life Taxonomy (OTT) [9]. We found that SILVA, RDP and Greengenes can be mapped into NCBI and OTT with few conflicts, but not vice versa. There is a great deal of difference between taxonomies that arise because of the differences in size and structure. Taxonomic classifications Each of the five taxonomies that we compare is based on a mixture of sources that have been compiled into taxonomies in different ways. They differ in both size and resolution (see Table 1). All taxonomies assign ranks to their nodes, the seven main ones being domain, phylum, class, order, family, genus and species. However, RDP and SILVA only go down to the genus level, whereas NCBI and OTT go down to the species level and below. The two latter taxonomies also have a number of intermediate ranks and contain many intermediate nodes (Fig. 2 a). To simplify the comparison of taxonomies, we will consider only nodes associated with the seven main ranks. Fig. 2Composition of the five taxonomies. a Composition by rank type. Main rank stands for either root, domain, phylum, class, order, family, genus or species; intermediate includes all ‘sub–’, ‘infra–’, ‘super–’ etc. ranks. b Composition with respect to the number of nodes at each rank from root to genus. Square areas correspond to the number of nodes at each rank in each taxonomic classification Table 1Overview of five taxonomic classificationsTaxonomyTypeNo. of nodesLowest rankLatest releaseSILVAManual12,117GenusSep 2016RDPSemi6,128GenusSep 2016GreengenesAutomatic3,093SpeciesMay 2013NCBIManual1,522,150SpeciesTodaya OTTAutomatic2,627,066SpeciesSep 2016 aFor the analyses we have used NCBI taxonomy as published on 5th Oct 2016 Figure 2 a shows the percentage of nodes that are assigned to a main rank in each of the five taxonomies. We found that all taxonomies have 1–2% of nodes with an intermediate rank (‘sub–’, ‘super–’ and other), except for Greengenes. Nodes with no rank assignment are found only in OTT (3.3%) and NCBI (13.3%). The latter taxonomic classification has the lowest percentage (84.4%) of nodes that fall into the category of main ranks. Figure 2 b shows the composition of the five taxonomies at all ranks down to the level of genus. The NCBI taxonomy has 2.7 times fewer genera and 1.9 times fewer species (not shown) than the OTT. In the following we describe each of the five taxonomies in more detail (summarized in Table 1). SILVA The SILVA database [4] contains taxonomic information for the domains of Bacteria, Archaea and Eukarya. It is based primarily on phylogenies for small subunit rRNAs (16S for prokaryotes and 18S for Eukarya). Taxonomic rank information for Archaea and Bacteria is obtained from Bergey’s Taxonomic Outlines [10–13] and from the List of Prokaryotic Names with Standing in Nomenclature (LPSN) [14], whereas eukaryotic taxonomy is based on the consensus views of the International Society of Protistologists [15, 16]. Taxonomic rank assignments in the SILVA database are manually curated [4]. For the comparisons we used the taxonomy associated with SILVA small subunit ribosomal RNAs (16S/18S) v128 as released on 29/09/2016. Ribosomal database project (RDP) The RDP database [3] is based on 16S rRNA sequences from Bacteria, Archaea and Fungi (Eukarya). It contains 16S rRNA sequences available from the International Nucleotide Sequence Database Collaboration (INSDC) [17] databases. Names of the organisms associated with the sequences are obtained as the most recently published synonym from Bacterial Nomenclature Up-to-Date [18]. Information on taxonomic classification for Bacteria and Archaea is based on the taxonomic roadmaps by Bergey’s Trust [19] and LPSN [14]. Taxonomic information for fungi is obtained from a hand-made classification dedicated to fungal taxonomy [3]. For the comparisons we used a taxonomy associated with RDP database of 16S rRNA (Bacteria and Archaea) and 28S rRNA (Fungi) sequences as released on 30/09/2016 (release 11.5). Greengenes (GG) The Greengenes taxonomy [6] is dedicated to Bacteria and Archaea. Classification is based on automatic de novo tree construction and rank mapping from other taxonomy sources (mainly NCBI). Phylogenetic tree is constructed from 16S rRNA sequences that have been obtained from public databases and passed a quality filtering. Sequences are aligned by their characters and secondary structure and then subjected to tree construction with FastTree [20]. Inner nodes are automatically assigned taxonomic ranks from NCBI supplemented with previous version of Greengenes taxonomy and CyanoDB [21]. For the comparisons we used a taxonomy associated with the Greengenes database as released on May 2013. Although Greengenes is still included in some metagenomic analyses packages, for example QIIME [22], it has not been updated for the last three years. NCBI The NCBI taxonomy [7] contains the names of all organisms associated with submissions to the NCBI sequence databases. It is manually curated based on current systematic literature, and uses over 150 sources, for example, the Catalog of Life [23], the Encyclopedia of Life [24], NameBank [25] and WikiSpecies [26] as well as some specific databases dedicated to particular groups of organisms. It contains some duplicate names that represent different organisms. Each node has a scientific name