jshoyer / ictvxlsx Goto Github PK

My goal here is to make it a little bit easier to use tables from the ICTV (the International Committee on the Taxonomy of Viruses) in R.

Depending on your goal, you might not need to use R at all. If you want to quickly browse the taxonomic hierarchy from the top down, the ICTV site includes a nice visual browser for current and past versions. Another interactive view that is cross-referenced by annotated host distribution is at https://www.ncbi.nlm.nih.gov/labs/virus (NCBI = US National Center for Biotechnology Information). The NCBI Taxonomy database provides a text-based hierarchical browser that can be used to access many types of data for each taxon. A separate interface at https://www.ncbi.nlm.nih.gov/datasets allows one to browse the NCBI taxonomic hierarchy while searching for multiple taxon names at once. As noted by Siddell et al. (2023) in a paper on the role of the ICTV, the NCBI Taxonomy team uses ICTV releases but there is a short lag each time, so it can be worthwhile to cross-reference the ICTV tables.

The main product that the ICTV distributes is an Excel file “species list” table (the MSL). Let's say that you have downloaded the current Excel table from ictv.global/msl (download shortcut https://ictv.global/msl/current) and put it in your current working directory.

library(readxl)

msl <-
  read_excel("ICTV_Master_Species_List_2023_MSL39.v1.xlsx",
             sheet = 2, .name_repair = "universal")

The ICTV distributes a second table, the metadata resource (ictv.global/vmr, shortcut https://ictv.global/vmr/current). This table includes recommended common names and abbreviations, which are distinct from formal taxonomic names. It also list an exemplar sequence (GenBank accession numbers) for each species and in some cases one or more additional (A) sequence rows. It can be useful to get just the exemplar rows.

library(dplyr)

vmr <-
  read_excel("VMR_MSL39_v1.xlsx",
             sheet = 1, .name_repair = "universal")

vmr_e <- filter(vmr,
                Exemplar.or.additional.isolate == "E") |>
  select(-Exemplar.or.additional.isolate)

Accession numbers for each segment for multipartite viruses can be split with regular expressions. For example:

library(stringr)

b_refseq_regex <- "DNA-B: NC_\\d+"
a_refseq_regex <- "DNA-A: NC_\\d+"
b_regex <- "DNA-B: [A-Z]+\\d+"
a_regex <- "DNA-A: [A-Z]+\\d+"

begomo <- filter(vmr, Genus == "Begomovirus") |>
    mutate(b_genbank = str_extract(Virus.GENBANK.accession, b_regex) |>
               substr(8, 999),
           genbank = str_extract(Virus.GENBANK.accession, a_regex) |>
               substr(8, 999),
           b_refseq = str_extract(Virus.REFSEQ.accession,
                                  b_refseq_regex) |>
               substr(8, 999),
           refseq = str_extract(Virus.REFSEQ.accession,
                                a_refseq_regex) |>
               substr(8, 999),
	   .keep = "unused")

Much higher-rank information is duplicated in the MSL and VMR, so I like to combine the tables. In effect this adds the exemplar sequence (e) information to the MSL. (If you never expect to use the revision history in the MSL, you might find it simpler to just use VMR alone.)

Join tables like so:

msl_e <- left_join(msl, vmr_e,
                   by = c("Realm", "Subrealm", "Kingdom", "Subkingdom",
                          "Phylum", "Subphylum", "Class", "Subclass",
                          "Order", "Suborder", "Family", "Subfamily",
                          "Genus", "Subgenus", "Species",
                          "Genome.Composition" = "Genome.composition"))

It is often convenient to add NCBI taxon IDs to (subsets of) this data frame. The best way I have found to do this so far is to navigate the taxonomy database and/or query it with a URL like so https://www.ncbi.nlm.nih.gov/taxonomy/?term=txid10239[Subtree] (possibly replacing 10239, the taxonomy ID covering all viruses, with a more specific ID). By clicking on ‘Send to’ one can export both a Taxon name file and a Taxid list file. Reading these files into R, cbinding them into a data frame, and then joining that data frame to the one above by species names (or genus names, or the names for any other rank of interest) is straightforward.

Others have considered using one or more R packages to access the full history of the taxonomy, including taxon merging, splitting, reassignment-moves, and renaming (taxa package issue 136 and linked taxize issue 695).

If one wants to access the taxonomy from any directory, it can be worthwhile to wrap it in a package, storing either the Excel files themselves or R objects derived from them. A package might be especially useful if one is comparing multiple versions of the tables, which necessitates more specific/verbose variable names. I am exploring this idea on branches in this repository. Feel free to file an issue if this is of interest.