INDRA v1.18.0

Package structure and dependencies

• Starting with this release, INDRA supports Python 3.6+ only.
• The live curation service for the World Modelers use case is now in a separate repository at https://github.com/indralab/indra_wm_service.
• New dependencies for this release include obonet, tqdm and pybiopax.
• Pyjnius is no longer a necessary requirement for any INDRA module (though it can still be optionally used for local JAR-based reading with Eidos and Reach).

Core assembly modules

• A major change in the current release is the introduction of the indra.ontology module which replaces the indra.preassembler.hierarchy_manager module, for more details, see #1100.
• New BioOntology at indra.ontology.bio which implements naming, identifier mapping and hierarchical relationships among 2.1 million relevant entities (up from 250k in the previous release). The BioOntology also subsumes the old cellular component hierarchy, activity hierarchy, and modification hierarchy, which has now been removed.
• New WorldOntology at indra.ontology.world which replaces multiple reader-specific ontologies for the WorldModelers use case.
• New ontology web service in indra.ontology.app which allows using the IndraOntology as a service for ID mapping, isa/partof queries, name lookups, etc.
• New VirtualOntology indra.ontology.virtual which acts as a client for the ontology web service.
• The name standardization and grounding standardization code has been reimplemented to use the IndraOntology, and is now under the indra.ontology.standardize module with significantly extended mapping capabilities, and many new functions.
• The Preassembler has been adapted to use the IndraOntology and is extended with new options in indra.preassembler.
• The indra.preassembler.ontology_mapper module along with a host of other scripts for making hierarchies (all replaced by the new IndraOntology) have been deprecated and removed.
• Improved grounding mapping and disambiguation logic when multiple texts are available for an agent in indra.preassembler.grounding_mapper.
• Many extensions to grounding mapping resources in indra.resources.grounding.
• Bug fixes in indra.pipeline for handling some corner cases.

Knowledge sources

• With the relase of PyBioPAX (https://github.com/indralab/pybiopax), the BioPAX processor has been fully reimplemented, and is now a pure-Python module that no longer requires pyjnius to operate. See indra.sources.biopax.
• New API and processor for the Comparative Toxicogenomics Database (CTD) at indra.sources.ctd.
• Improved extraction of grounding from PyBEL models in indra.sources.bel.
• New API and processor for the DrugBank database at indra.sources.drugbank.
• New API and processor for the OmniPath web service at indra.sources.omnipath.
• New API and processor for Biofactoid at indra.sources.biofactoid.
• Overhauled API and processor for the Target Affinity Spectrum in indra.sources.tas with significantly extended content and improved naming and grounding.
• The BEL RDF processor has been deprecated and removed with the PyBEL processor continuing to serve as the entrypoint for processing BEL content in indra.sources.bel.
• Multiple new biology-oriented endpoints in the Eidos API in indra.sources.eidos.
• The CWMS RDF processor has been removed in factor of the EKB/XML processor which continues to operate at indra.sources.cwms.

Model assemblers

• Extended support for knowledge sources and simplified support for evidence count and source count annotations in the HtmlAssembler in indra.assemblers.html. There are also several bug fixes in handling some statement types in this module, as well as improved links out to literature content.
• Extended support for knowledge source annotations in the CX assembler in indra.assemblers.cx.
• Handling negative modification conditions in the EnglishAssembler indra.assemblers.english.
• Improved signed graph assembly from BEL graphs in indra.assemblers.pybel.

Model analysis / explanation

• New dedicated module for causal path finding methods in indra.explanation.pathfinding.
• Generalized and improved statement checking in all submodules of indra.explanation.model_checker
  with support for open-ended search.

Resources and database clients

• New Taxonomy client in indra.databases.taxonomy_client
• New DrugBank client in indra.databases.drugbank_client
• Reimplemented ChEBI client based on OBO in indra.databases.chebi_client.
• Extended ChEBML client with naming information in indra.databases.chembl_client.
• Extended LINCS client with more naming and mapping information in indra.databases.lincs_client.
• The BioGRID client has been removed from indra.databases.biogrid_client since it has become redundant with indra.sources.biogrid.
• Support for the PubChem JSON web service endpoint in indra.databases.pubchem_client.
• The network relevance client has been removed from indra.databases.relevance_client since it relied on a web service that is no longer operational.
• Many extensions to mapping and naming resources, in particular for chemicals.

Literature clients

• New client for the DART document infrastructure used in the World Modelers use case in indra.literature.dart.
• Better search by MeSH terms in the PubMed client in indra.literature.pubmed_client.

REST API

• The INDRA REST API has been completely reimplemented in a new framework, and has been extended significantly in its functionalities. It also comes with improved Swagger documentation with a public instance
  running at http://api.indra.bio:8000.