Indra

Improvements in:
- BioPAX API/Processor default extractions
- PysbAssembler default parameterization
- Handling mutations in reading, refinements and assembly

New/improved natural language model examples in:
- `models/hello_indra`
- `models/p53_model`
- `models/braf_model`
- `models/ras_pathway`

Bug fixes and improvements in
- CyJS, CX, English and PySB Assemblers
- BioPAX, REACH and TRIPS Processors
- GeneNetwork tool and the SiteMapper
- The RAS Machine

Representation features:
- RegulateAmount (to represent synthesis/transcription, degradation) INDRA Statements collected from inputs and assembled into models
- RegulateActivity INDRA Statements now generalize activation/inhibition
- New subclasses of Modification INDRA Statements
- Bioentities extended and is mapped to in INDRA input processors

Input processors:
- BEL processor extracts indirect Statements
- New API and processor for Sparser NLP system in `indra.sparser`
- Various improvements in all existing input processors

Output assemblers:
- Cytoscape JS assembler
- PySB assembler generalized and extended with PysbPreassembler
- Various extensions in all existing output assemblers

Core assembly modules:
- Improved and generalized BeliefEngine implementation in `indra.belief`
- Improved, generalized and extended MechLinker implementation in `indra.mechlinker`
- Optimized Preassembler in `indra.preassembler` with multiprocessing option

Tools:
- Assemble corpus tool in `indra.tools.assemble_corpus` exposes all assembly functionalities and adds many Statement filters
- Model Checker in `indra.tools.model_checker` verifies executable models with respect to observations represented as indirect INDRA Statements
- Expand family tool in `indra.tools.expand_families` using Bioentities relationships
- Improved high-throughput reading tools in `indra.tools.reading`

Other:
- REST API exposing main INDRA functionalities as a web service
- New example models in `models`
- Extended documentation and tutorials in `doc`

New features:
- Python 3 support in addition to maintaining compatibility with Python 2
- Universal handling of unicode within INDRA
- Bioentities added as a submodule as a basis for entity hierarchies (for protein family and complex relationships)
- Resources for performing high-throughput literature reading (`indra.tools.reading`) with Amazon cluster support
- Belief Engine (`indra.belief`) applies belief propagation based on evidence from multiple sources to score the believability of INDRA Statements
- Grounding Mapper (`indra.preassembler.grounding_mapper`) fixes named entity grounding to databases based on a use case specific mapping table
- New output assemblers added: SIF assembler (`indra.assemblers.sif_assembler`), Index Card assembler (`indra.assemblers.index_card_assembler`)
- Index Card processor as an input source (`indra.index_cards`)
- Several benchmarks are now available in `indra.benchmarks`

Documentation now available at http:// http://indra.readthedocs.io/

New features:
- Refactored assemblers module (`indra.assemblers`) containing PySB, CX, English, IndexCard, Graph and SBGN assemblers
- Literature module (`indra.literature`) with PubMed, PMC, Elsevier and CrossRef clients
- Mechanism linker (`indra.mechlinker`) to simplify and infer missing links between mechanisms
- Generalized representation for active forms of proteins (`indra.statements.ActiveForm`) and activation events (`indra.statements.Activation`)
- Representation for Agent cellular location and translocation processes (`indra.statements.Translocation`)
- Relevance service from NDEx based on network heat diffusion
- JSON serialization and deserialization of INDRA Statements (`indra.statements.Statement.to_json`)

New tools:
- Ras Machine (`models/rasmachine`) - a framework for building incrementally updated use case-specific models based on a prior network and new literature as it appears
- Incremental model (`indra.tools.incremental_model`) - a class for assembling a model incrementally as new mechanisms become available
- Gene network (`indra.tools.gene_network`) - a tool for extracting and assembling a network of known mechanisms given a gene list of interest from the PathwayCommons database and the BEL Large Corpus
- Executable subnetwork (`indra.tools.executable_subnetwork`) - a tool for extracting a subnetwork of limited scope from a large set of INDRA Statements and instantiating it as a rule-based executable model

New features:
- Site mapper for the preassembler
- CX assembler for visualization in Cytoscape and NDEx

Fixes:
- Resource files are now within the module
- Installation via setup.py now excludes jnius, which has to be manually installed
- Some fixes in the BioPAX processor