Skip to content

Builds lists of Taxonomy Identifiers based on the NCBI Taxonomy Database

License

Notifications You must be signed in to change notification settings

ebi-jdispatcher/taxonomy-resolver

Repository files navigation

PyPI license PyPI version

Taxonomy Resolver builds an NCBI Taxonomy Tree structure based on the NCBI Taxonomy Database classification. Taxonomy Resolver can be used to validate Taxonomy Identifiers (TaxIDs) against the Tree or to generate lists of TaxIDs, based on some TaxID of interest (e.g. higher-level rank (node) in the tree).

The main features of Taxonomy Resolver are:

  1. Downloading taxonomy dump files from the NCBI ftp server
  2. Building an NCBI Taxonomy Tree data structure based on the NCBI Taxonomy classification
  3. Writing and loading the Tree structure in pickle format
  4. Building a slimmer "filtered" Tree (based on a list of TaxIDs) to improve performance
  5. Quick lookup to see if a TaxID exists in the Tree (i.e. is valid)
  6. Generate lists of all children TaxIDs that compose a particular Node (sub-tree)
  7. Generate lists of children TaxIDs based on a list of included and excluded TaxIDs (included and excluded sub-trees)
  8. Filtering the resulting list of children TaxIDs, for example to cleanup TaxIDs that are not observed in a dataset of interest

Taxonomy Resolver initially builds a tree hierarchy structure resulting in deeply nested dictionaries. To retrieve a full tree or a sub-tree, a lot of iteration takes place, following the path from the node of interest down the hierarchy. This approach does not scale well, especially for very large trees. Thus, in Taxonomy Resolver, the tree is represented following a different approach, commonly referred to as the Nested Set Model. In the Nested Set Model, we can look at a tree hierarchy differently, not as connected nodes, but as nested containers. The nested set model is a particular technique for representing nested sets in relational databases, which we implement here in a pandas DataFrame. For that, the full tree is traversed with Modified Preorder Tree Traversal strategy. In a preorder traversal, the root node is visited first, then recursively a preorder traversal of the left sub-tree, followed by a recursive preorder traversal of the right subtree, in order, until every node has been visited. The modified strategy allows us to capture the 'left' and 'right' (lft and rgt, respectively) boundaries of each nested container. Querying and searching is much faster with this approach because finding a subtree is as simple as filtering/searching for the nodes where lft > Node's lft and rgt < Node's rgt. Likewise, find the full path to a node is as simple as filtering/searching for the nodes where lft < Node's lft and rgt > Node's rgt.

Both traversal markup (left and right), depth and node indexes are captured for each node in the tree. The following Taxonomy Tree is used as a test mock tree - see nodes_mock.dmp:

testdata/mock_tree_diagram_50.png

The resulting tree can be represented in tabular form:

index tax_id parent_id rank depth lft rgt
1 1 1 root 1 1 58
2 2 1 rank 2 2 2 57
3 3 2 rank 3 3 3 28
4 5 3 rank 4 4 4 21
5 10 5 rank 5 5 5 8
6 19 10 rank 6 6 6 7
7 11 5 rank 5 5 9 10
8 12 5 rank 5 5 11 20
9 20 12 rank 6 6 12 13
10 21 12 rank 6 6 14 19
11 25 21 rank 7 7 15 16
12 26 21 rank 7 7 17 18
13 6 3 rank 4 4 22 23
14 7 3 rank 4 4 24 27
15 13 7 rank 5 5 25 26
16 4 2 rank 3 3 29 56
17 8 4 rank 4 4 30 39
18 14 8 rank 5 5 31 36
19 22 14 rank 6 6 32 33
20 23 14 rank 6 6 34 35
21 15 8 rank 5 5 37 38
22 9 4 rank 4 4 40 55
23 16 9 rank 5 5 41 42
24 17 9 rank 5 5 43 44
25 18 9 rank 5 5 45 54
26 24 18 rank 6 6 46 53
27 27 24 rank 7 7 47 52
28 28 27 rank 8 8 48 49
29 29 27 rank 8 8 50 51

Installation requires Python 3.10+ (recommended version 3.11). Additional requirements, which will be downloaded and installed automatically. See full list of dependencies in requirements.txt

Dependencies for the Python tools developed here, are the typical Python stack (3.10+ and pip). A good approach is to set a virtual environment:

virtualenv -p `which python` env
source ./env/bin/activate
pip install --upgrade -r requirements.txt
pip freeze > requirements_from_freeze.txt
deactivate

Taxonomy Resolver can be installed with pip:

pip install taxonomy-resolver

Taxonomy Resolver can be used as a Python module or via the CLI provided.

Example of typical usage of the Taxonomy Resolver module is provided below:

from taxonomyresolver import TaxonResolver

resolver = TaxonResolver()

# Download the NCBI Taxonomy Data Dump
dumpfile = "taxdmp.zip"
resolver.download(dumpfile, "zip")

# Building the NCBI Taxonomy Tree data structure
resolver.build(dumpfile)

# Saving the Tree data structure as Pickle format
treefile = "tree.pickle"
resolver.write(treefile, "pickle")

# Get a list of children TaxIDs that compose a set of TaxIDs
searchfile = "taxids_search.txt"
tax_ids = resolver.search(searchfile)

# Write the TaxIDs to a file
taxidsfile = "taxids_list.txt"
with open(outfile, "w") as outfile:
    outfile.write("\n".join(tax_ids))

When a Taxonomy Tree is already available one can simply load it with resolver.load():

from taxonomyresolver import TaxonResolver

resolver = TaxonResolver()

# Loading the NCBI Taxonomy Tree
treefile = "tree.pickle"
resolver.load(treefile, "pickle")

# Validate a set of TaxIDs against the Tree data structure
validatefile = "taxids_validate.txt"
valid = resolver.validate(validatefile)
if valid:
  print(f"TaxIDs in {validatefile} are valid!")

Explore the CLI by running taxonomy-resolver (COMMAND) --help

Usage: taxonomy-resolver [OPTIONS] COMMAND1 [ARGS]... [COMMAND2
                            [ARGS]...]...

  Taxonomy Resolver: Build a NCBI Taxonomy Tree, validate and search TaxIDs.

Options:
  --version   Show the version and exit.
  -h, --help  Show this message and exit.

Commands:
  build     Build a NCBI Taxonomy Tree data structure.
  download  Download the NCBI Taxonomy dump file ('taxdmp.zip').
  search    Searches a Tree data structure and writes a list of TaxIDs.
  validate  Validates a list of TaxIDs against a Tree data structure.

Additional help is provided for each command, for example, running taxonomy-resolver (command) --help, returns:

Usage: taxonomy-resolver search [OPTIONS]

  Searches a Tree data structure and writes a list of TaxIDs.

Options:
  -in, --infile TEXT             Path to input NCBI BLAST dump or a prebuilt tree file, (currently: 'pickle').  [required]
  -out, --outfile TEXT           Path to output file.
  -inf, --informat TEXT          Input format (currently: 'pickle').
  -taxid, --taxid TEXT           Comma-separated TaxIDs or pass multiple values. Output to STDOUT by default, unless an output file is provided.
  -taxids, --taxidinclude TEXT   Path to Taxonomy id list file used to search the Tree.
  -taxidexc, --taxidexc TEXT     Comma-separated TaxIDs or pass multiple values.
  -taxidse, --taxidexclude TEXT  Path to Taxonomy id list file excluded from the search.
  -taxidsf, --taxidfilter TEXT   Path to Taxonomy id list file used to filter the search.
  -ignore, --ignoreinvalid       Ignores invalid TaxIDs.
  -level, --log_level TEXT       Log level to use. Expects: 'DEBUG', 'INFO', 'WARN', 'ERROR', and 'CRITICAL'.
  -l, --log_output TEXT          File name to be used to writing logging.
  --quiet                        Disables logging.
  -sep, --sep TEXT               String Separator to use.
  -indx, --indx INTEGER          String positional index to use (starts with 0).
  -h, --help                     Show this message and exit

Getting the NCBI Taxonomy Data from the NCBI ftp server:

taxonomy-resolver download -out taxdmp.zip

Building a Tree structure from the taxdmp.zip file and saving it in JSON (or alternatively in pickle format):

taxonomy-resolver build -in taxdmp.zip -out tree.pickle

Filtering an existing Tree structure in pickle format by passing a file containing a list of TaxIDs, and saving it in pickle format:

taxonomy-resolver build -in tree.pickle -inf pickle -out tree_filtered.pickle -outf pickle -taxidf testdata/taxids_filter.txt

Load a previously built Tree data structure in pickle format and generating a list of TaxIDs that compose the hierarchy based on list of TaxIDs:

taxonomy-resolver search -in tree.pickle -taxids testdata/taxids_search.txt

Load a previously built Tree data structure in pickle format and generating a list of TaxIDs (included TaxIDs), exclude TaxIDs from the search (excluded TaxIDs), and filter the final result to only those TaxIDs that are available in the list of filter TaxIDs (filtered TaxIDs):

taxonomy-resolver search -in tree.pickle -taxids testdata/taxids_search.txt -taxidse testdata/taxids_exclude.txt -taxidsf testdata/taxids_filter.txt -out taxids_list.txt

Validating a list of TaxIDs against a Tree data structure in pickle format:

taxonomy-resolver validate -in tree.pickle -taxids testdata/taxids_validate.txt

See CONTRIBUTING.rst for more information about contributing to Taxonomy Resolver.

If you find any bugs or issues please log them in the issue tracker.

See release notes on CHANGELOG.rst

I would like to thank Adrian Tivey for insightful discussions.

The European Bioinformatics Institute - EMBL-EBI, is an Intergovernmental Organization which, as part of the European Molecular Biology Laboratory family, focuses on research and services in bioinformatics.

Apache License 2.0. See license for details.

About

Builds lists of Taxonomy Identifiers based on the NCBI Taxonomy Database

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published