IEDB Protein Tree

James A. Overton [email protected]
Rebecca Tauber [email protected]

The IEDB Molecule Finder includes all species and ancestors from the organism tree, plus proteins from their reference proteomes.

Status

This code is a revised version of the protein tree, designed to replace the legacy protein tree code.

It generates two different products:

• protein-tree.owl.gz proteins used in IEDB linked to their NCBITaxon organism species
• molecule-tree.owl.gz protein tree plus non-peptide tree

Requirements

• a Unix system (Linux, macOS)
• GNU Make 3.81+
• Python 3.6 with the rdflib package
• Java 8 for running ROBOT

Usage

To make all products, run:

make all

This will generate the gzipped versions of both trees, and then clean up the intermediate build files.

To create the protein and molecule trees without removing the intermediate build files, use:

make trees

Protein and Molecule Trees

The process generates the protein tree from various tabular inputs and merges with the legacy non-peptide tree to create the molecule tree. The necessary dependencies that must be manually added are:

• dependencies/parent_protein.tsv assignments of proteins referenced in the IEDB and their parent proteomes
  • Used to generate dependencides/parent-proteins.csv
• dependencies/source_parent.tsv assignments of all sources to reference proteins from the reference proteomes
  • Used to generate dependencies/source-parents.csv
• dependencies/proteomes.tsv assignments of proteome species to their proteome IDs

The other dependencies are automatically retrieved with curl (force update by deleting):

• dependencies/organism-tree.owl nodes for all taxa used by the IEDB
• dependencies/subspecies-tree.owl organism tree plus all ranks used by the IEDB
• dependencies/non-peptide-tree.owl non-peptide molecular entities

Intermediate Products

All products here are generated in the temp directory.

• organism-proteins.ttl all classes from organism-tree.owl as proteins
• upper.ttl top-level structure for proteins including 'protein' and 'material entity'
• source-synonyms.ttl synonyms as annotations from source-parents.csv
• iedb-proteins.ttl proteins from parent-proteins.csv as subclasses of their species protein
• ncbi-classes.tsv all NCBITaxon classes used by IEDB proteins as proteome IDs
• included-classes.tsv table of NCBITaxon species included in organism-proteins.ttl
• missing-classes.txt list of NCBITaxon species NOT included in organism-proteins.ttl, but used in iebd-proteins.ttl
• taxon-proteins.owl missing subspecies (in missing-classes.txt) filtered from subspecies-tree.owl
• merged.owl combination of taxon-proteins.owl, upper.ttl, iedb-proteins.ttl, source-synonyms.ttl, and branches.owl.gz (see below)

Branches

The full branches.owl.gz file contains all species protein branches. These are build from their UniProt reference proteomes. This process works with any species that has a reference protein, specified by dependencies/proteomes.tsv. This file is merged into the protein tree to include all details about a species proteome.

Each time a new parent-protein table is used to run a build (dependencies/parent_protein.tsv), a new proteome for a species in the organism tree will be fetched only if its proteins in the parent-protein table have changed. If these proteins have changed, an updated reference proteome will be downloaded from UniProt and used to rebuild the branch node for that species.

After the build is over, the dependencies/parent-proteins.csv file is copied to dependencies/parent-proteins-last.csv. If this file does not exist, all proteomes will be re-downloaded. When a new parent-proteins table is generated or added, it is compared to the -last version to find differences in proteins.