ImmuneDB is a system to facilitate efficient storage and analysis of high-throughput B- and T-cell sequence data. It provides V- and J-gene identification, clonal assignment, lineage construction, selection pressure calculation, and thorough exporting functionality.
It also provides an intuitive and useful web interface a demo of which you can see here.
The system aims to:
Reduce ad-hoc scripting: Data analysis performed on an ad-hoc basis with custom scripts and data formats is error-prone and leads to inconsistencies. ImmuneDB provides a standardized analysis platform, performing many common tasks automatically.
Minimize flat-files: Flat files are currently the standard method of data exchange in the biological sciences. There are a myriad of drawbacks when using these including a lack of referential integrity, unclear provenance, and non-standardized formats.
ImmuneDB attempts to reduce the need for flat files, through the use of an industry-leading database, MySQL. When data must be exchanged as a flat-file, many export options, including FASTA and tab-delineation, are available.
Interoperate with existing tools: ImmuneDB integrates tools from other researchers to provide features such as lineage construction, genotyping, and selection pressure calculations. Further, ImmuneDB can import and export in a variety of common formats, making it compatible with the larger AIRR ecosystem of tools.
ImmuneDB has four primary components:
Clonal identification pipeline: A series of steps which identify probable V- and J-genes from raw FASTA files and subsequently groups sequences from the same subject into clones. At each step, statistics and metadata are stored in the underlying database for later analysis.
Analysis framework: A module that provide detailed analysis of sequences. This includes mutation analysis, gene utilization breakdowns, and feature distributions (e.g. CDR3 length, V-gene utilization).
API: ImmuneDB includes two APIs. First, the REST API allows native and web applications (via AJAX) to interact with the underlying database without writing any database specific code. Further, the REST API is language agnostic, using HTTP to exchange data.
Second, the Python API which implements the data models and querying functionality. This allows developers to write customized queries directly interacting with the database.
Web interface: Distributed separately of ImmuneDB is a web interface which utilizes the REST API to allow for easy browsing of analysis results.
Each of the components can be used independently so long as the underlying database is properly populated.
- MySQL: The database (agnostic of implementation like MariaDB) which enforces rigorous data guarantees, indexes the data for fast retrieval, and enforces atomicity of operations
- SQLAlchemy: The interface between Python and the database which abstracts SQL queries from the implementation and enforces the data model within Python.
- Bottle: A lightweight web-framework for Python which serves content via the REST API allowing web-applications to issue AJAX requests to the ImmuneDB framework.