taxize.Rmd

---
title: Introduction to taxize
author: Scott Chamberlain
date: "2020-09-17"
output: rmarkdown::html_vignette
vignette: >
    %\VignetteIndexEntry{Introduction to taxize}
    %\VignetteEngine{knitr::rmarkdown}
    %\VignetteEncoding{UTF-8}
---



`taxize` is a taxonomic toolbelt for R. `taxize` wraps APIs for a large suite of taxonomic databases availab on the web.

## Installation

First, install and load `taxize` into the R session.


```r
install.packages("taxize")
```


```r
library("taxize")
```

Advanced users can also download and install the latest development copy from GitHub (https://github.com/ropensci/taxize)

## Resolve taxonomic name

This is a common task in biology. We often have a list of species names and we want to know a) if we have the most up to date names, b) if our names are spelled correctly, and c) the scientific name for a common name. One way to resolve names is via the Global Names Resolver (GNR) service provided by the Encyclopedia of Life. Here, we are searching for two misspelled names:


```r
temp <- gnr_resolve(c("Helianthos annus", "Homo saapiens"))
head(temp)
#> # A tibble: 6 x 5
#>   user_supplied_name submitted_name  matched_name     data_source_title    score
#>   <chr>              <chr>           <chr>            <chr>                <dbl>
#> 1 Helianthos annus   Helianthos ann… Helianthus annus uBio NameBank         0.75
#> 2 Helianthos annus   Helianthos ann… Helianthus annu… Catalogue of Life     0.75
#> 3 Helianthos annus   Helianthos ann… Helianthus annu… ITIS                  0.75
#> 4 Helianthos annus   Helianthos ann… Helianthus annu… NCBI                  0.75
#> 5 Helianthos annus   Helianthos ann… Helianthus annu… GRIN Taxonomy for P…  0.75
#> 6 Helianthos annus   Helianthos ann… Helianthus annu… Union 4               0.75
```

The correct spellings are *Helianthus annuus* and *Homo sapiens*.

taxize takes the approach that the user should be able to make decisions about what resource to trust, rather than making the decision. The GNR service provides data from a variety of data sources. The user may trust a specific data source, thus may want to use the names from that data source. In the future, we may provide the ability for taxize to suggest the best match from a variety of sources.

Another common use case is when there are many synonyms for a species. In this example, we have three synonyms of the currently accepted name for a species.


```r
mynames <- c("Helianthus annuus ssp. jaegeri", "Helianthus annuus ssp. lenticularis", "Helianthus annuus ssp. texanus")
(tsn <- get_tsn(mynames, accepted = FALSE))
══  3 queries  ═══════════════
✔  Found:  Helianthus annuus ssp. jaegeri
✔  Found:  Helianthus annuus ssp. lenticularis
✔  Found:  Helianthus annuus ssp. texanus
══  Results  ═════════════════

● Total: 3 
● Found: 3 
● Not Found: 0
[1] "525928" "525929" "525930"
attr(,"class")
[1] "tsn"
attr(,"match")
[1] "found" "found" "found"
attr(,"multiple_matches")
[1] FALSE FALSE FALSE
attr(,"pattern_match")
[1] FALSE FALSE FALSE
attr(,"uri")
[1] "https://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=525928"
[2] "https://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=525929"
[3] "https://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=525930"
lapply(tsn, itis_acceptname)
[[1]]
  submittedtsn      acceptedname acceptedtsn author
1       525928 Helianthus annuus       36616     L.

[[2]]
  submittedtsn      acceptedname acceptedtsn author
1       525929 Helianthus annuus       36616     L.

[[3]]
  submittedtsn      acceptedname acceptedtsn author
1       525930 Helianthus annuus       36616     L.
```

## Retrieve higher taxonomic names

Another task biologists often face is getting higher taxonomic names for a taxa list. Having the higher taxonomy allows you to put into context the relationships of your species list. For example, you may find out that species A and species B are in Family C, which may lead to some interesting insight, as opposed to not knowing that Species A and B are closely related. This also makes it easy to aggregate/standardize data to a specific taxonomic level (e.g., family level) or to match data to other databases with different taxonomic resolution (e.g., trait databases).

A number of data sources in taxize provide the capability to retrieve higher taxonomic names, but we will highlight two of the more useful ones: Integrated Taxonomic Information System (ITIS) and National Center for Biotechnology Information (NCBI). First, we'll search for two species, *Abies procera} and *Pinus contorta* within ITIS.


```r
specieslist <- c("Abies procera","Pinus contorta")
classification(specieslist, db = 'itis')
#> ══  2 queries  ═══════════════
#> ✔  Found:  Abies procera
#> ✔  Found:  Pinus contorta
#> ══  Results  ═════════════════
#> 
#> ● Total: 2 
#> ● Found: 2 
#> ● Not Found: 0
#> $`Abies procera`
#>               name          rank     id
#> 1          Plantae       kingdom 202422
#> 2    Viridiplantae    subkingdom 954898
#> 3     Streptophyta  infrakingdom 846494
#> 4      Embryophyta superdivision 954900
#> 5     Tracheophyta      division 846496
#> 6  Spermatophytina   subdivision 846504
#> 7        Pinopsida         class 500009
#> 8          Pinidae      subclass 954916
#> 9          Pinales         order 500028
#> 10        Pinaceae        family  18030
#> 11           Abies         genus  18031
#> 12   Abies procera       species 181835
#> 
#> $`Pinus contorta`
#>               name          rank     id
#> 1          Plantae       kingdom 202422
#> 2    Viridiplantae    subkingdom 954898
#> 3     Streptophyta  infrakingdom 846494
#> 4      Embryophyta superdivision 954900
#> 5     Tracheophyta      division 846496
#> 6  Spermatophytina   subdivision 846504
#> 7        Pinopsida         class 500009
#> 8          Pinidae      subclass 954916
#> 9          Pinales         order 500028
#> 10        Pinaceae        family  18030
#> 11           Pinus         genus  18035
#> 12  Pinus contorta       species 183327
#> 
#> attr(,"class")
#> [1] "classification"
#> attr(,"db")
#> [1] "itis"
```

It turns out both species are in the family Pinaceae. You can also get this type of information from the NCBI by doing `classification(specieslist, db = 'ncbi')`.

Instead of a full classification, you may only want a single name, say a family name for your species of interest. The function `tax_name` is built just for this purpose. As with the `classification` function you can specify the data source with the `db` argument, either ITIS or NCBI.


```r
tax_name("Helianthus annuus", get = "family", db = "ncbi")
#> ══  1 queries  ═══════════════
#> ✔  Found:  Helianthus+annuus
#> ══  Results  ═════════════════
#> 
#> ● Total: 1 
#> ● Found: 1 
#> ● Not Found: 0
#>     db             query     family
#> 1 ncbi Helianthus annuus Asteraceae
```

It may happen that a data source does not provide information on the queried species, than one could take the result from another source and union the results from the different sources.

## Interactive name selection

As mentioned most databases use a numeric code to reference a species. A general workflow in taxize is: Retrieve Code for the queried species and then use this code to query more data/information.

Below are a few examples. When you run these examples in R, you are presented with a command prompt asking for the row that contains the name you would like back; that output is not printed below for brevity. In this example, the search term has many matches. The function returns a data frame of the matches, and asks for the user to input what row number to accept.


```r
get_uid("Pinus")
#> ══  1 queries  ═══════════════
#> ✔  Found:  Pinus
#> ══  Results  ═════════════════
#> 
#> ● Total: 1 
#> ● Found: 1 
#> ● Not Found: 0
#> [1] "3337"
#> attr(,"class")
#> [1] "uid"
#> attr(,"match")
#> [1] "found"
#> attr(,"multiple_matches")
#> [1] FALSE
#> attr(,"pattern_match")
#> [1] FALSE
#> attr(,"uri")
#> [1] "https://www.ncbi.nlm.nih.gov/taxonomy/3337"
```

In another example, you can pass in a long character vector of taxonomic names (although this one is rather short for demo purposes):


```r
splist <- c("annona cherimola", 'annona muricata', "quercus robur")
get_tsn(splist, searchtype = "scientific")
#> ══  3 queries  ═══════════════
#> ✔  Found:  annona cherimola
#> ✔  Found:  annona muricata
#> ✔  Found:  quercus robur
#> ══  Results  ═════════════════
#> 
#> ● Total: 3 
#> ● Found: 3 
#> ● Not Found: 0
#> [1] "506198" "18098"  "19405" 
#> attr(,"class")
#> [1] "tsn"
#> attr(,"match")
#> [1] "found" "found" "found"
#> attr(,"multiple_matches")
#> [1] FALSE FALSE  TRUE
#> attr(,"pattern_match")
#> [1] FALSE FALSE  TRUE
#> attr(,"uri")
#> [1] "https://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=506198"
#> [2] "https://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=18098" 
#> [3] "https://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=19405"
```

There are functions for many other sources

* `get_boldid()`
* `get_eolid()`
* `get_gbifid()`
* `get_nbnid()`
* `get_tpsid()`

Sometimes with these functions you get a lot of data back. In these cases you may want to limit your choices. Soon we will incorporate the ability to filter using `regex` to limit matches, but for now, we have a new parameter, `rows`, which lets you select certain rows. For example, you can select the first row of each given name, which means there is no interactive component:


```r
get_nbnid(c("Zootoca vivipara","Pinus contorta"), rows = 1)
#> ══  2 queries  ═══════════════
#> ✔  Found:  Zootoca vivipara
#> ✔  Found:  Pinus contorta
#> ══  Results  ═════════════════
#> 
#> ● Total: 2 
#> ● Found: 2 
#> ● Not Found: 0
#> [1] "NHMSYS0001706186" "NBNSYS0000004786"
#> attr(,"class")
#> [1] "nbnid"
#> attr(,"match")
#> [1] "found" "found"
#> attr(,"multiple_matches")
#> [1] TRUE TRUE
#> attr(,"pattern_match")
#> [1] FALSE FALSE
#> attr(,"uri")
#> [1] "https://species.nbnatlas.org/species/NHMSYS0001706186"
#> [2] "https://species.nbnatlas.org/species/NBNSYS0000004786"
```

Or you can select a range of rows


```r
get_nbnid(c("Zootoca vivipara","Pinus contorta"), rows = 1:3)
#> ══  2 queries  ═══════════════
#> ✔  Found:  Zootoca vivipara
#> ✔  Found:  Pinus contorta
#> ══  Results  ═════════════════
#> 
#> ● Total: 2 
#> ● Found: 2 
#> ● Not Found: 0
#> [1] "NHMSYS0001706186" "NBNSYS0000004786"
#> attr(,"class")
#> [1] "nbnid"
#> attr(,"match")
#> [1] "found" "found"
#> attr(,"multiple_matches")
#> [1] TRUE TRUE
#> attr(,"pattern_match")
#> [1] TRUE TRUE
#> attr(,"uri")
#> [1] "https://species.nbnatlas.org/species/NHMSYS0001706186"
#> [2] "https://species.nbnatlas.org/species/NBNSYS0000004786"
```

In addition, in case you don't want to do interactive name selection in the case where there are a lot of names, you can get all data back with functions of the form, e.g., `get_tsn_()`, and likewise for other data sources. For example:


```r
out <- get_nbnid_("Poa annua")
NROW(out$`Poa annua`)
#> [1] 25
```

That's a lot of data, so we can get only certain rows back


```r
get_nbnid_("Poa annua", rows = 1:10)
#> $`Poa annua`
#>                guid     scientificName    rank taxonomicStatus
#> 1  NBNSYS0000002544          Poa annua species        accepted
#> 2  NBNSYS0200001901       Bellis annua species        accepted
#> 3  NBNSYS0200003392   Triumfetta annua species        accepted
#> 4  NBNSYS0200002555        Lonas annua species        accepted
#> 5  NHMSYS0000456951  Carrichtera annua species        accepted
#> 6  NHMSYS0000461807 Poa labillardierei species        accepted
#> 7  NHMSYS0000461808      Poa ligularis species        accepted
#> 8  NHMSYS0000461817     Poa sieberiana species        accepted
#> 9  NHMSYS0000461805         Poa gunnii species        accepted
#> 10 NHMSYS0000461801     Poa costiniana species        accepted
```

## Coerce numerics/alphanumerics to taxon IDs

We've also introduced in `v0.5` the ability to coerce numerics and alphanumerics to taxonomic ID classes that are usually only retrieved via `get_*()` functions.

For example, adfafd


```r
as.gbifid(get_gbifid("Poa annua")) # already a uid, returns the same
#> ══  1 queries  ═══════════════
#>    gbifid             scientificname    rank   status matchtype
#> 1 2704179               Poa annua L. species ACCEPTED     EXACT
#> 2 8422205 Poa annua Cham. & Schltdl. species  SYNONYM     EXACT
#> 3 7730008           Poa annua Steud. species DOUBTFUL     EXACT
#> ✖  Not Found:  Poa annua
#> ══  Results  ═════════════════
#> 
#> ● Total: 1 
#> ● Found: 0 
#> ● Not Found: 1
#> [1] NA
#> attr(,"class")
#> [1] "gbifid"
#> attr(,"match")
#> [1] "not found"
#> attr(,"multiple_matches")
#> [1] TRUE
#> attr(,"pattern_match")
#> [1] FALSE
as.gbifid(2704179) # numeric
#> [1] "2704179"
#> attr(,"class")
#> [1] "gbifid"
#> attr(,"match")
#> [1] "found"
#> attr(,"multiple_matches")
#> [1] FALSE
#> attr(,"pattern_match")
#> [1] FALSE
#> attr(,"uri")
#> [1] "https://www.gbif.org/species/2704179"
as.gbifid("2704179") # character
#> [1] "2704179"
#> attr(,"class")
#> [1] "gbifid"
#> attr(,"match")
#> [1] "found"
#> attr(,"multiple_matches")
#> [1] FALSE
#> attr(,"pattern_match")
#> [1] FALSE
#> attr(,"uri")
#> [1] "https://www.gbif.org/species/2704179"
as.gbifid(list("2704179","2435099","3171445")) # list, either numeric or character
#> [1] "2704179" "2435099" "3171445"
#> attr(,"class")
#> [1] "gbifid"
#> attr(,"match")
#> [1] "found" "found" "found"
#> attr(,"multiple_matches")
#> [1] FALSE FALSE FALSE
#> attr(,"pattern_match")
#> [1] FALSE FALSE FALSE
#> attr(,"uri")
#> [1] "https://www.gbif.org/species/2704179"
#> [2] "https://www.gbif.org/species/2435099"
#> [3] "https://www.gbif.org/species/3171445"
```

These `as.*()` functions do a quick check of the web resource to make sure it's a real ID. However, you can turn this check off, making this coercion much faster:


```r
system.time( replicate(3, as.gbifid(c("2704179","2435099","3171445"), check=TRUE)) )
#>    user  system elapsed 
#>   0.092   0.003   4.850
system.time( replicate(3, as.gbifid(c("2704179","2435099","3171445"), check=FALSE)) )
#>    user  system elapsed 
#>   0.002   0.000   0.002
```

## What taxa are downstream of my taxon of interest?

If someone is not a taxonomic specialist on a particular taxon he likely does not know what children taxa are within a family, or within a genus. This task becomes especially unwieldy when there are a large number of taxa downstream. You can of course go to a website like Wikispecies or Encyclopedia of Life to get downstream names. However, taxize provides an easy way to programatically search for downstream taxa for the Integrated Taxonomic Information System.


```r
apis_itis_id <- 154395 # id for Apis, fetched beforehand to save time here
downstream(apis_itis_id, downto = "species", db = "itis")
#> $`154395`
#>       tsn parentname parenttsn rankname          taxonname rankid
#> 1 1128092       Apis    154395  species     Apis laboriosa    220
#> 2  154396       Apis    154395  species     Apis mellifera    220
#> 3  763550       Apis    154395  species Apis andreniformis    220
#> 4  763551       Apis    154395  species        Apis cerana    220
#> 5  763552       Apis    154395  species       Apis dorsata    220
#> 6  763553       Apis    154395  species        Apis florea    220
#> 7  763554       Apis    154395  species Apis koschevnikovi    220
#> 8  763555       Apis    154395  species   Apis nigrocincta    220
#> 
#> attr(,"class")
#> [1] "downstream"
#> attr(,"db")
#> [1] "itis"
```

## Direct children

You may sometimes only want the direct children. We got you covered on that front, with methods for ITIS and NCBI.

The direct children (genera in this case) of _Pinaceae_ using NCBI data:


```r
children("Pinaceae", db = "ncbi")
#> $Pinaceae
#>    childtaxa_id childtaxa_name childtaxa_rank
#> 1        123600     Nothotsuga          genus
#> 2         64685        Cathaya          genus
#> 3          3358          Tsuga          genus
#> 4          3356    Pseudotsuga          genus
#> 5          3354    Pseudolarix          genus
#> 6          3337          Pinus          genus
#> 7          3328          Picea          genus
#> 8          3325          Larix          genus
#> 9          3323     Keteleeria          genus
#> 10         3321         Cedrus          genus
#> 11         3319          Abies          genus
#> 
#> attr(,"class")
#> [1] "children"
#> attr(,"db")
#> [1] "ncbi"
```

## Get NCBI ID from GenBank Ids

With accession numbers


```r
genbank2uid(id = 'AJ748748')
#> [[1]]
#> [1] "282199"
#> attr(,"class")
#> [1] "uid"
#> attr(,"match")
#> [1] "found"
#> attr(,"multiple_matches")
#> [1] FALSE
#> attr(,"pattern_match")
#> [1] FALSE
#> attr(,"uri")
#> [1] "https://www.ncbi.nlm.nih.gov/taxonomy/282199"
#> attr(,"name")
#> [1] "Nereida ignava 16S rRNA gene, type strain 2SM4T"
```

With gi numbers


```r
genbank2uid(id = 62689767)
#> [[1]]
#> [1] "282199"
#> attr(,"class")
#> [1] "uid"
#> attr(,"match")
#> [1] "found"
#> attr(,"multiple_matches")
#> [1] FALSE
#> attr(,"pattern_match")
#> [1] FALSE
#> attr(,"uri")
#> [1] "https://www.ncbi.nlm.nih.gov/taxonomy/282199"
#> attr(,"name")
#> [1] "Nereida ignava 16S rRNA gene, type strain 2SM4T"
```

## Matching species tables with different taxonomic resolution

Biologist often need to match different sets of data tied to species. For example, trait-based approaches are a promising tool in ecology. One problem is that abundance data must be matched with trait databases. These two data tables may contain species information on different taxonomic levels and possibly data must be aggregated to a joint taxonomic level, so that the data can be merged. taxize can help in this data-cleaning step, providing a reproducible workflow:

We can use the mentioned `classification`-function to retrieve the taxonomic hierarchy and then search the hierarchies up- and downwards for matches. Here is an example to match a species with names on three different taxonomic levels.


```r
A <- "gammarus roeseli"

B1 <- "gammarus roeseli"
B2 <- "gammarus"
B3 <- "gammaridae"

A_clas <- classification(A, db = 'ncbi')
#> ══  1 queries  ═══════════════
#> ✔  Found:  gammarus+roeseli
#> ══  Results  ═════════════════
#> 
#> ● Total: 1 
#> ● Found: 1 
#> ● Not Found: 0
B1_clas <- classification(B1, db = 'ncbi')
#> ══  1 queries  ═══════════════
#> ✔  Found:  gammarus+roeseli
#> ══  Results  ═════════════════
#> 
#> ● Total: 1 
#> ● Found: 1 
#> ● Not Found: 0
B2_clas <- classification(B2, db = 'ncbi')
#> ══  1 queries  ═══════════════
#> ✔  Found:  gammarus
#> ══  Results  ═════════════════
#> 
#> ● Total: 1 
#> ● Found: 1 
#> ● Not Found: 0
B3_clas <- classification(B3, db = 'ncbi')
#> ══  1 queries  ═══════════════
#> ✔  Found:  gammaridae
#> ══  Results  ═════════════════
#> 
#> ● Total: 1 
#> ● Found: 1 
#> ● Not Found: 0

B1[match(A, B1)]
#> [1] "gammarus roeseli"
A_clas[[1]]$rank[tolower(A_clas[[1]]$name) %in% B2]
#> [1] "genus"
A_clas[[1]]$rank[tolower(A_clas[[1]]$name) %in% B3]
#> [1] "family"
```

If we find a direct match (here *Gammarus roeseli*), we are lucky. But we can also match Gammaridae with *Gammarus roeseli*, but on a lower taxonomic level. A more comprehensive and realistic example (matching a trait table with an abundance table) is given in the vignette on matching.