Loading phylogenetic data into R

Martin R. Smith martin.smith@durham.ac.uk

2020-06-08

It can be a bit fiddly to get a phylogenetic dataset into R, particularly if you are not used to working with files in the Nexus format.

First off, make sure that you are comfortable telling R where to find a file.

Then you are ready to load raw data:

From an Excel spreadsheet

If your data is in an Excel spreadsheet, one way to load it into R is using the xlsx package. First you’ll have to install it:

install.packages('xlsx') # You only need to do this once

Then you should prepare your Excel spreadsheet such that each row corresponds to a taxon, and each column to a character.

Then you can read the data from the Excel file by telling R which sheet, rows and columns contain your data:

library('xlsx')
raw_data <- as.matrix(read.xlsx(filename, 
                      sheetIndex = 1, # Loads sheet number 1 from the excel file
                      rowIndex = 2:21, # Extracts rows 2 to 21
                      colIndex = 2:26, # Extracts columns B to Z
                      header = FALSE
                      ))

# In this example, the names of taxa are in column 1
taxon_names <- read.xlsx(filename, sheetIndex = 1, rowIndex = 2:21, 
                         colIndex = 1, as.data.frame=FALSE)
                         
rownames(raw_data) <- taxon_names

From a Nexus file

TreeTools contains an inbuilt Nexus parser:

raw_data <- ReadCharacters(filename)
# Or, to go straight to PhyDat format:
as_phydat <- ReadAsPhyDat(filename)

This will extract character names and codings from a dataset. It’s been written to work with datasets downloaded from MorphoBank, but my aim is for this function to handle most valid (and many invalid) NEXUS files. If you find a file that this function can’t handle, please let me know and I’ll try to fix it.

In the meantime, alternative Nexus parsers are available: try

raw_data <- ape::read.nexus.data(filename)

Non-standard elements of a Nexus file might be beyond the capabilities of ape’s parser. In particular, you will need to replace spaces in taxon names with an underscore, and to arrange all data into a single block starting BEGIN DATA. You’ll need to strip out comments, character definitions and separate taxon blocks.

The function readNexus in package phylobase uses the NCL library and promises to be more powerful, but I’ve not been able to get it to work.

From a TNT file

A TNT format dataset downloaded from MorphoBank can be parsed with ReadTntCharacters, which might also handle other TNT-compatible files. If there’s a file that’s not being read correctly, please let me know and I’ll try to fix it.

raw_data <- ReadTntCharacters(filename)
# Or, to go straight to PhyDat format:
my_data <- ReadTntAsPhyDat(filename)

Processing raw data

The next stage is to get the raw data into a format that most R packages can understand. If you’ve used the ReadAsPhyDat or ReadTntAsPhyDat functions, then you can skip this step – you’re already there.

Otherwise, you can try

my_data <- PhyDat(raw_data)

or if that doesn’t work,

my_data <- MatrixToPhyDat(raw_data)

These functions are pretty robust, but might return an error when they encounter an unexpected dataset format – if they don’t work on your dataset, please
let me know.

Failing that, you can enlist the help of the ‘phangorn’ package, which was installed when you installed ‘TreeTools’:

library('phangorn')
my_data <- phyDat(raw_data, type='USER', levels=c(0:9, '-'))

type='USER' tells the parser to expect morphological data.

The levels parameter simply lists all the states that any character might take. 0:9 includes all the integer digits from 0 to 9. If you have inapplicable data in your matrix, you should list - as a separate level as it represents an additional state (as handled by the Morphy implementation of (Brazeau, Guillerme, & Smith, 2019)). If you have more complicated ambiguities, you may need to use a contrast matrix to decode your matrix.

A contrast matrix translates the tokens used in your dataset to the character states to which they correspond: for example decoding ‘A’ to {01}. For more details, see the ‘phangorn-specials’ vignette in the phangorn package, accessible by typing ‘?phangorn’ in the R prompt and navigating to index > package vignettes.

contrast.matrix <- matrix(data=c(
# 0 1 -  # Each column corresponds to a character-state
  1,0,0, # Each row corresponds to a token, here 0, denoting the 
         # character-state set {0} 
  0,1,0, # 1 | {1}
  0,0,1, # - | {-}
  1,1,0, # A | {01}
  1,1,0, # + | {01}
  1,1,1  # ? | {01-}
), ncol=3, # ncol should correspond to the number of columns in the matrix
byrow=TRUE); 
dimnames(contrast.matrix) <- list(
  c(0, 1, '-', 'A', '+', '?'), # A list of the tokens corresponding to each row
                               # in the contrast matrix
  c(0, 1, '-') # A list of the character-states corresponding to the columns 
               # in the contrast matrix
)

contrast.matrix
##   0 1 -
## 0 1 0 0
## 1 0 1 0
## - 0 0 1
## A 1 1 0
## + 1 1 0
## ? 1 1 1

If you need to use a contrast matrix, convert the data using

my.phyDat <- phyDat(my.data, type='USER', contrast=contrast.matrix)

What next?

You might want to:

References

Brazeau, M. D., Guillerme, T., & Smith, M. R. (2019). An algorithm for morphological phylogenetic analysis with inapplicable data. Systematic Biology, 68(4), 619–631. doi:10.1093/sysbio/syy083

Faith, D. P., & Trueman, J. W. H. (2001). Towards an inclusive philosophy for phylogenetic inference. Systematic Biology, 50(3), 331–350. doi:10.1080/10635150118627