Combining data from OToL and other sources.

One of the major goals of rotl is to help users combine data from other sources with the phylogenetic trees in the Open Tree database. This examples demonstrates how a user might connect data they have collected to trees from Open Tree.

Get Open Tree ids to match your data.

Let’s say you have a dataset where each row represents a measurement taken from one species, and your goal is to put these measurements in some phylogenetic context. Here’s a small example: the best estimate of the mutation rate for a set of unicellular Eukaryotes along with some other property of those species which might explain the mutation rate:

csv_path <- system.file("extdata", "protist_mutation_rates.csv", package = "rotl")
mu <- read.csv(csv_path, stringsAsFactors=FALSE)
mu
##                     species       mu pop.size genome.size
## 1   Tetrahymena thermophila 7.61e-12 1.12e+08    1.04e+08
## 2    Paramecium tetraurelia 1.94e-11 1.24e+08    7.20e+07
## 3 Chlamydomonas reinhardtii 2.08e-10 1.00e+08    1.12e+08
## 4  Dictyostelium discoideum 2.90e-11 7.40e+06    3.40e+07
## 5  Saccharomyces cerevisiae 3.30e-10 1.00e+08    1.25e+08
## 6       Saccharomyces pombe 2.00e-10 1.00e+07    1.25e+08

If we want to get a tree for these species we need to start by finding the unique ID for each of these species in the Open Tree database. We can use the Taxonomic Name Resolution Service (tnrs) functions to do this. Before we do that we should see if any of the taxonomic contexts, which can be used to narrow a search and avoid conflicts between different codes, apply to our group of species:

library(rotl)
tnrs_contexts()
## Possible contexts:
##    Animals 
##       Birds, Tetrapods, Mammals, Amphibians, Vertebrates, Arthropods 
##       Arthropods, Molluscs, Nematodes, Platyhelminthes, Annelids, Cnidarians 
##       Cnidarians, Arachnides, Insects 
##    Bacteria 
##       SAR group, Archaea, Excavata, Amoebae, Centrohelida, Haptophyta 
##       Haptophyta, Apusozoa, Diatoms, Ciliates, Forams 
##    Fungi 
##       Basidiomycetes, Ascomycetes 
##    Land plants 
##       Hornworts, Mosses, Liverworts, Vascular plants, Club mosses, Ferns 
##       Ferns, Seed plants, Flowering plants, Monocots, Eudicots, Rosids 
##       Rosids, Asterids, Asterales, Asteraceae, Aster, Symphyotrichum 
##       Symphyotrichum, Campanulaceae, Lobelia 
##    All life

Hmm, none of those groups contain all of our species. In this case we can search using the All life context and the function tnrs_match_names:

taxon_search <- tnrs_match_names(names=mu$species, context_name="All life")
knitr::kable(taxon_search)
search_string unique_name approximate_match ott_id is_synonym is_deprecated number_matches
tetrahymena thermophila Tetrahymena thermophila FALSE 180195 FALSE FALSE 1
paramecium tetraurelia Paramecium tetraurelia FALSE 568130 FALSE FALSE 1
chlamydomonas reinhardtii Chlamydomonas reinhardtii FALSE 33153 FALSE FALSE 1
dictyostelium discoideum Dictyostelium discoideum FALSE 160850 FALSE FALSE 1
saccharomyces cerevisiae Saccharomyces cerevisiae FALSE 908549 FALSE FALSE 1
saccharomyces pombe Schizosaccharomyces pombe FALSE 990004 TRUE FALSE 1

Good, all of our species are known to Open Tree. Note, though, that one of the names is a synonym. Saccharomyces pombe is older name for what is now called Schizosaccharomyces pombe. As the name suggests, the Taxonomic Name Resolution Service is designed to deal with these problems (and similar ones like misspellings), but it is always a good idea to check the results of tnrs_match_names closely to ensure the results are what you expect.

In this case we have a good ID for each of our species so we can move on. Before we do that, let’s ensure we can match up our original data to the Open Tree names and IDs by adding them to our data.frame:

mu$ott_name <- taxon_search$unique_name
mu$ott_id <- taxon_search$ott_id

Find a tree with your taxa

Now let’s find a tree. There are two possible options here: we can search for published studies that include our taxa or we can use the ‘synthetic tree’ from Open Tree. We can try both approaches.

Published trees

Before we can search for published studies or trees, we should check out the list of properties we can use to perform such searches:

studies_properties()
## $tree_properties
##  [1] "ot:treebaseOTUId"           "ot:nodeLabelMode"          
##  [3] "ot:originalLabel"           "oti_tree_id"               
##  [5] "ot:ottTaxonName"            "ot:inferenceMethod"        
##  [7] "ot:tag"                     "ot:comment"                
##  [9] "ot:treebaseTreeId"          "ot:branchLengthDescription"
## [11] "ot:treeModified"            "ot:studyId"                
## [13] "ot:branchLengthTimeUnits"   "ot:ottId"                  
## [15] "is_deprecated"              "ot:branchLengthMode"       
## [17] "ot:treeLastEdited"          "ot:nodeLabelDescription"   
## 
## $study_properties
##  [1] "ot:studyModified"             "ot:focalClade"               
##  [3] "ot:focalCladeOTTTaxonName"    "ot:focalCladeOTTId"          
##  [5] "ot:studyPublication"          "ot:studyLastEditor"          
##  [7] "ot:tag"                       "ot:focalCladeTaxonName"      
##  [9] "ot:studyLabel"                "ot:comment"                  
## [11] "ot:authorContributed"         "ot:studyPublicationReference"
## [13] "ot:curatorName"               "ot:studyId"                  
## [15] "ot:studyYear"                 "ot:studyUploaded"            
## [17] "is_deprecated"                "ot:dataDeposit"

We have ottIds for our taxa, so let’s use those IDs to search for trees that contain them. Starting with our first species Tetrahymena thermophila we can use studies_find_trees to do this search.

studies_find_trees(property="ot:ottId", value="180195")
## [1] study_ids       dat             n_matched_trees match_tree_ids 
## <0 rows> (or 0-length row.names)

Well… that’s not very promising. We can repeat that process for all of the IDs to see if the other species are better represented.

hits <- lapply(mu$ott_id, studies_find_trees, property="ot:ottId", detailed = FALSE)
sapply(hits, function(x) sum(x[["n_matched_trees"]]))
## [1]  0  0  0  0 32  3

OK, most of our species are not in any of the published trees available. You can help fix this sort of problem by making sure you submit your published trees to Open Tree.

A part of the synthesis tree

Thankfully, we can still use the complete Tree of Life made from the combined results of all of the published trees and taxonomies that go into Open Tree. The function tol_induced_subtree will fetch a tree relating a set of IDs.

Using the default arguments you can get a tree object into your R session:

tr <- tol_induced_subtree(ott_ids=mu$ott_id)
plot(tr)

Connect your data to the tips of your tree

Now we have a tree for of our species, how can we use the tree and the data together?

The package phylobase provide an object class called phylo4d, which is designed to represent a phylogeny and data associated with its tips. In oder to get our tree and data into one of these objects we have to make sure the labels in the tree and in our data match exactly. That’s not quite the case at the moment (tree labels have underscores and IDs appended):

mu$ott_name[1]
## [1] "Tetrahymena thermophila"
tr$tip.label[4]
## [1] "Tetrahymena_thermophila_ott180195"

We can use sub to the remove the underscores and ottId from the tree (check out ?regex to see how these patterns work) and %in% to confirm that each of the modified labels matches a taxon in our data.frame:

tr$tip.label <- sub("_ott\\d+", "", tr$tip.label)
tr$tip.label <- sub("_", " ", tr$tip.label)
tr$tip.label %in% mu$ott_name
## [1] TRUE TRUE TRUE TRUE TRUE TRUE

Ok, now the tips are together we can make a new dataset. The phylo4d() functions matches tip lables to the row names of a data.frame, so let’s make a new dataset that contains just the relevant data and has row names to match the tree

library(phylobase)
mu_numeric <- mu[,c("mu", "pop.size", "genome.size")]
rownames(mu_numeric) <- mu$ott_name
tree_data <- phylo4d(tr, mu_numeric)

And now we can plot the data and the tree together

plot(tree_data)

What next

The demonstration get’s you to the point of visualizing your data in a phylogenetic context. But there’s a lot more you do with this sort of data in R. For instance, you could use packages like ape, caper, phytools and mcmcGLMM to perform phylogenetic comparative analyses of your data. You could gather more data on your species using packages that connect to trait databases like rfishbase, AntWeb or rnpn which provides data from the US National Phenology Network. You could also use rentrez to find genetic data for each of your species, and use that data to generate branch lengths for the phylogeny.