Biolink API

An API providing access to information on biologically and biomedically relevant entities, and the relationships between them, including:

• genes, gene products, proteins
• diseases, phenotypes, traits and clinical measurements
• pathways, biological process
• substances: small molecules, drugs, chemical entities
• biological and molecular roles and activities
• genotypes, alleles, sequence variants; for plants, germplasms
• environmental contexts and exposures
• individual organisms: patients, cohorts, model organisms
• cell lines and cell types
• investigations: experiments, clinical trials and 'natural experiments'
• genomic features
• phylogenies
• metadata: publications, ontology terms, database metadata, prefixes

This repository provides an example server for the biolink API. The swagger is also generated from this.

Demo

This API is designed to be implemented or partially implemented via a variety of databases.

The Monarch instance provides access to a wide variety of aggregated data:

http://api.monarchinitiative.org/api/

Datamodel

See the swagger UI for more details. Click on 'model' under any of the routes.

The primary abstraction used in the modeling is the distinction between named objects and associations.

• Named objects include things like genes, drugs, pathways.
  • specific types subclass from a more generic type
• Associations connect these, usually via some evidence and provenance information.
  • Some associations can be direct, others are indirect or inferred
  • Where associations are inferred, these is a graph of evidence tracing the primary associations

Examples

See EXAMPLE-QUERIES.md

These examples are compiled from the behave tests

Writing client code

The API uses Swagger, which means you can take advantage of a variety of swagger tooling

For example, to generate a python client:

swagger-codegen generate -i http://api.monarchinitiative.org/api/swagger.json -l $* -o $@

Contributing to BioLink-API

This is the repo:

https://github.com/monarch-initiative/biolink-api/

You can run a server instance locally with very little effort (less than one minute), see below:

Setting up

pyvenv venv
source venv/bin/activate
pip install -r requirements.txt
export PYTHONPATH=.:$PYTHONPATH

to run:

python biolink/app.py

Then look at:

http://localhost:5000/api/

For the swagger docs

Note that only a small subset has been implemented

Organization

This repo contains multiple sub-packages. These can all be used in python programs independent of biolink or a web service context.

• biolink FlaskREST service implementation for BioLink API
• biomodel Data access objects
• biogolr Python API onto Monarch and GO golr instances
• scigraph Python API onto Monarch SciGraph instance
• prefixcommons Python code for working with ID prefixes

Goals

This API will wrap and integrate number of different more modular APIs and database engines or analysis services. The idea is that the API implementation will do the right thing - for example, using Solr for searches but injecting results with fast in-memory traversal of ontology graphs.

This is a proof of concept implementation. May be implemented using a JVM language, e.g. scala in future.

Example API calls (OLD)

All assocations, first 10:

http://localhost:5000/api/link/search/

All mouse gene-phenotype associations:

http://localhost:5000/api/link/search/gene/phenotype/?map_identifiers=MGI&subject_taxon=NCBITaxon:10090

Same but with IDs mapped from NCBIGene to MGI (see #5):

http://localhost:5000/api/link/search/gene/phenotype/?subject_taxon=NCBITaxon:10090

phenotypes for a given gene

http://localhost:5000/api/bio/gene/ZFIN:ZDB-GENE-050417-357/phenotypes/

GO terms for a given gene (uses GO golr)

http://localhost:5000/api/bio/gene/ZFIN:ZDB-GENE-050417-357/function/

Query association by ID:

http://localhost:5000/api/link/cfef92b7-bfa3-44c2-a537-579078d2de37

Evidence graph as bbop-graph:

http://localhost:5000/api/evidence/graph/cfef92b7-bfa3-44c2-a537-579078d2de37

Evidence graph as image:

http://localhost:5000/api/evidence/graph/cfef92b7-bfa3-44c2-a537-579078d2de37/image

Anatomical entities for a given phenotype:

http://localhost:5000/api/bioentity/phenotype/ZP:0004204/anatomy/

Implementation and Project Organization

This is intended as a think wrapper layer, integrating existing services, as shown here:

This is only very partially integrated

We have some business logic in the following sub-packages:

• scigraph - python API for Monarch Neo4J instance and generic SciGraph functions
• biogolr - python API for any golr instance (GO or Monarch)
• obographs - python ontology object model and utilities. See obographs
• causalmodels - python API for LEGO, and wrapper to GO triplestore
• prefixcommons - python utilities for handling prefixes/CURIEs

These may eventually migrate to their own repos, see #16

Solr indices

Most of the link subset is implemented via the Monarch Golr, with function queries coming from the GO instance. See:

monarch-initiative#14

SciGraph

Two different ways of accessing this are used:

• Direct access via BOLT/Cypher
• Access to SciGraph API layer

Triplestore

The lego calls are imlemented as calls to the GO triplestore.

OwlSim

http://owlsim3.monarchinitiative.org/api/docs/

Notes