Turn biomedical knowledge into searchable, evidence-backed relationships
Access a biomedical knowledge graph built from published research and use it in practical research workflows. IKraph API helps researchers and developers search direct relationships, explore indirect multi-hop paths, and retrieve literature-backed evidence across diseases, genes, drugs, chemicals, pathways, and more.
IKraph is designed for teams that want more than raw graph data. The API supports direct relation search for focused biological questions and indirect relation search for path discovery and hypothesis generation. Results are backed by literature evidence and can be retrieved efficiently through a result-based workflow built for reuse, filtering, and downstream analysis.
Why researchers use IKraph API
Search direct biological relationships
Run 1-hop searches across biomedical entity types such as Disease, Gene, Drug, Chemical, Pathway, Anatomy, Biological Process, Molecular Function, and more. Filter by relationship type, direction, correlation, novelty, date range, and disease subtype expansion.
Explore indirect paths for hypothesis generation
Search indirect relationships between two entities through intermediate nodes. The API supports path discovery with configurable intermediate nodes, optional direct-relation inclusion, novelty filtering, date filtering, and evidence refinement. This makes it useful for mechanism exploration and literature-driven hypothesis generation.
Retrieve evidence-backed results
Use lightweight result retrieval when you want evidence counts only, or retrieve full evidence details when you need PMID-level support, extracted sentences, confidence scores, and publication dates. Optional LLM-based refinement can validate evidences before results are returned.
Built for real workflows
IKraph is not just an endpoint. It is the foundation for a broader tools ecosystem that includes a Python client, CLI support, analytics helpers, visualization, Jupyter integrations, and domain-specific pipelines for use cases such as drug repurposing, biomarker discovery, and hypothesis generation.
Key features
Direct relation search
Search for known 1-hop relationships between biomedical entities using graph traversal and evidence filtering. This is ideal for questions such as disease-gene associations, drug-pathway relationships, or chemical-disease links.
Indirect relation search
Discover multi-hop paths between two entities through intermediate nodes. The current API supports indirect searches with configurable intermediate node count and returns ordered paths with node and relationship details.
Flexible filtering
Refine results by relationship type, direction, correlation type, novelty, publication date range, subtype expansion, probability method, and optional direct relation inclusion for indirect queries.
Evidence details when you need them
Use one retrieval endpoint for lightweight browsing with evidence counts only, and another for full evidence details including PMID, sentence text, entity spans, confidence score, and publication date.
Efficient result retrieval
Direct search stores results and returns a result_id, which can then be used to retrieve saved results. This separation supports efficient reuse of previous searches and lighter result browsing before pulling full evidence.
Tools ecosystem in development
The first tools release is centered on ikraph-client, followed by CLI, analytics, visualization, Jupyter, and pipeline packages. The goal is to make IKraph usable in scripts, notebooks, presentations, and reproducible research workflows.
From query to publication-ready evidence
01
Submit a search
For direct relation search, send a query with your entity types, entity names, and filters. The API runs the search, stores the result, and returns a result_id rather than sending back the full relation payload immediately.
02
Retrieve the level of detail you need
Use the result_id to retrieve lightweight results with evidence counts only, or request full evidence details for literature review and traceability.
03
Build on top of the results
Move results into downstream workflows such as scripting, notebooks, network analysis, visualization, or domain-specific pipelines. The tools roadmap is designed to support outputs such as pandas, JSON, CSV, TSV, and network graph formats.
A simple, transparent credit system
IKraph API is metered with credits so cost stays predictable. You see exactly what each call costs, retries on the same query are free, and unsuccessful calls do not consume your balance.
Pay only for what you call
Every endpoint charges a flat base fee per call. Details and evidence endpoints add a per-result fee on top, so lighter queries stay cheap and heavier ones scale predictably with the data they return.
Cost is visible in every response
Every successful response carries credits_consumed and remaining_credits, so your client can log spend, enforce budgets, and show usage to end users without guessing what a call cost.
Cache hits are free
Identical requests within a 5-minute per-user window are served from cache with credits_consumed of 0. Retries, double renders, and exploratory parameter tweaks do not burn credits.
Failed calls are not billed
If validation, authorization, or the upstream backend fails, the base charge is cancelled and your balance is restored. You only pay when the call actually delivers data.
Current rates are 10 credits per base call, 1 credit per returned relation, and 10 credits per returned evidence. Prices may change without notice — always trust the credits_consumed field on the response over any client-side estimate. See the full pricing rules →
Supported entity types
IKraph API supports a broad range of biomedical entity types, including Gene, Chemical, Disease, Pathway, Species, Anatomy, Biological Process, CellLine, Cellular Component, DNAMutation, Molecular Function, and Pharmacologic Class.
Who it is for
Biomedical researchers who want to explore relationships across disease, gene, drug, and pathway space
Bioinformatics and computational biology teams building custom workflows
Drug discovery teams interested in target discovery, repurposing, combination hypotheses, or biomarker discovery
Developers who want programmatic access to evidence-backed biomedical graph data
What is coming with the IKraph tools package
We are building IKraph for how researchers actually work. The tools roadmap starts with ikraph-client as the foundation. It is designed to wrap the API with high-level search functions, authentication handling, credit awareness, caching, retry logic, and flexible output formats. On top of that, the ecosystem will expand to include command-line tools, network analysis helpers, visualization tools, notebook integrations, and domain-specific research pipelines.
ikraph-client
Python-based data access through ikraph-client.
ikraph-cli
Command-line scripting through ikraph-cli.
ikraph-analytics
Network and path analysis through ikraph-analytics.
ikraph-visualize
Interactive and presentation-ready visualization through ikraph-visualize.
ikraph-jupyter
Notebook-based exploratory work through ikraph-jupyter.
ikraph-pipelines
Domain workflows such as drug repurposing, biomarker discovery, and literature-driven hypothesis generation through ikraph-pipelines.
Frequently asked questions
The API supports direct relation search for 1-hop relationships and indirect relation search for multi-hop paths between two entities. Direct searches are useful for focused lookups, while indirect searches are useful for mechanism exploration and hypothesis generation.
You can retrieve either lightweight results with evidence counts only or full evidence details. Full evidence results include PMID, extracted sentence text, entity spans, confidence score, and publication date.
Yes. The API supports publication date filtering and novelty filtering, along with direction, correlation, and relationship-type filters.
Yes. For disease-based direct searches, you can choose whether to search a single disease entity or expand into the MeSH disease sub-tree and retrieve grouped subtype results.
Yes. The indirect relation endpoint is specifically designed to help users explore paths between entities through intermediate nodes, which can support literature-driven hypothesis generation. The tools roadmap also includes hypothesis-generation pipelines.
The tools plan starts with ikraph-client as the core package. It is intended to provide a higher-level Python interface with request validation, output conversion, caching, credit monitoring, retry logic, and async support over time.
The tools plan includes support for outputs such as DataFrame, JSON, dict, CSV, TSV, and NetworkX graph representations so users do not have to manually parse raw responses.
The result_id-based design separates search from retrieval. This makes it easier to reuse saved results, browse lightweight summaries first, and retrieve full evidence only when needed.
IKraph API is metered with credits. Every endpoint charges a base fee per call, and details/evidence endpoints add a per-result fee on top. Every successful response includes credits_consumed and remaining_credits, identical requests within a 5-minute window are served from cache for free, and failed calls are not billed. The full pricing rules live in the Credit System section of the v3 API reference.
Ready to explore IKraph API?
If you are interested in using IKraph API for biomedical research, discovery workflows, or custom data applications, we would be happy to share access details and documentation.
IKraph API brings biomedical knowledge graph search, evidence-backed retrieval, and multi-hop discovery into practical research workflows. It is built for teams that want to move from raw data access to usable scientific workflows.