NCATS-Tangerine/kgx

Name: kgx

Owner: NCATS Data Translator Project - Tangerine Team

Description: knowledge graph exchange tools

Created: 2018-04-24 02:09:07.0

Updated: 2018-05-09 23:41:22.0

Pushed: 2018-05-09 23:41:23.0

Homepage:

Size: 82

Language: Python

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

knowledge graph interchange

A utility library and set of command line tools for exchanging data in knowledge graphs.

The tooling here is partly generic but intended primarily for building the translator-knowledge-graph.

For additional background see the Translator Knowledge Graph Drive

Installation

install -r requirements.txt
on setup.py install

Command Line Usage

Use the --help flag to get help. Right now there is a single command:

e: kgx dump [OPTIONS] [INPUT]... OUTPUT

ansforms a knowledge graph from one representation to another
PUT  : any number of files or endpoints
TPUT : the output file

ons:
input-type TEXT   Extention type of input files: ttl, json, csv, rq, tsv,
                  graphml
output-type TEXT  Extention type of output files: ttl, json, csv, rq, tsv,
                  graphml
help              Show this message and exit.

CSV/TSV representation require two files, one that represents the vertex set and one for the edge set. JSON, TTL, and GRAPHML files represent a whole graph in a single file. For this reason when creating CSV/TSV representation we will zip the resulting files in a .tar file.

The format will be inferred from the file extention. But if this cannot be done then the --input-type and --output-type flags are useful to tell the program what formats to use. Currently not all conversions are supported.

Here are some examples that mirror the tests:

x dump --output-type=csv tests/resources/x1n.csv tests/resources/x1e.csv target/x1out
 created at: target/x1out.tar
x dump tests/resources/x1n.csv tests/resources/x1e.csv target/x1n.graphml
 created at: target/x1n.graphml
x dump tests/resources/monarch/biogrid_test.ttl target/bgcopy.csv
 created at: target/bgcopy.csv.tar
x dump tests/resources/monarch/biogrid_test.ttl target/x1n.graphml
 created at: target/x1n.graphml
x dump tests/resources/monarch/biogrid_test.ttl target/x1n.json
 created at: target/x1n.json

Internal Representation

Internal representation is networkx MultiDiGraph which is a property graph.

The structure of this graph is expected to conform to the tr-kg standard, briefly summarized here:

• Nodes
  • id : required
  • name : string
  • category : string. broad high level type. Corresponds to label in neo4j
  • extensible other properties, depending on
• Edges
  • subject : required
  • predicate : required
  • object : required
  • extensible other fields

Serialization/Deserialization

Intended to support

• Generic Graph Formats
• local or remote files
  • CSV
  • TSV (such as the RKB adapted data loading formats)
  • RDF (Monarch/OBAN style, …)
  • GraphML
  • CX
• remote store via query API
  • neo4j/bolt
  • RDF

RDF

Neo4J

Neo4j implements property graphs out the box. However, some implementations use reification nodes. The transform should allow for de-reification.