Cytegeist: A framework for biological data analysis Adam Treister1, Anja Mirenska2, Christian Hennig2

1) Gladstone Institutes, San Francisco, CA 2) Zellkraftwerk GmbH, Hannover, Germany

A New Cellular Metaphor

Complex experimental analysis is broken down into individual

steps executed by individual pieces of software. The traditional

process of chaining analytical steps together is known as

pipelining. The plumbing metaphor is accepted and ubiquitous,

but also reflects the limitations of an industrial age, where

architectures are static and changes require intervention by skilled

laborers to fundamentally reconstruct the pipeline. We propose

here a new approach to combining tools in a cellular

metaphor. We claim this architecture is more appropriate to

modern (Big Data) programming problems, where small amounts

of work are performed in parallel by independent work cells.

This is not a new computer architecture. It is merely a rebranding

of existing technology to better fit the mental model of

the biologist. The Service Oriented Architecture (SOA)

approach to enterprise computing is well-accepted as a scalable

abstraction that can distribute problems across a sea of virtual

machines. Well-defined service contracts are required to move

data between services, analogous to pathway diagrams used to

explain biological mechanisms.

As with cellular processes, there is a strict distinction between

inside and outside the cell, as well as a nucleus of essential data

that is translated into pieces of output. Internal pathways are used

to:

• normalize data

• map terms and abbreviations

• visualize data sets

• track workflow protocols.

Intermediate results and internal regulatory mechanisms float in a

protected environment inside the cell. At the same time a wide

array of services can access external databases and ontologies,

interact with robots, outsource extended calculations, share data,

and publish reports. As select analyses mature, they can be

cloned to scale up to the limits of available resources.

Cytegeist is fundamentally written in Java 1.8+. It is platform-

agnostic, data type-independent, and open source. Through

the CyRest interface, it inherits state transfer access to Python,

Ruby and R scripting. Interface elements are coded in JavaFX for

desktop and JavaScript for web clients. Data interchange is

generally pushed through JSON, XML and RDF channels, but

specific formats are easy to plug into the extensible architecture.

Summary

We propose a framework of analysis tools that interact through a

service API resembling an antigen binding model. The goal is to

provide a context that captures a modern service architecture in a

format comfortable to the biologist. We are looking for partners in

all aspects of developing open methods of analysis

and visualization.

Contact at: adam.treister@gladstone.ucsf.edu

References

1. Wikipedia; Instruction Pipelining; https://en.wikipedia.org/wiki/Instruction_pipelining. 2. Oracle; An Enterprise Architect’s Guide to Big Data; http://www.oracle.com/technetwork/topics/entarch/articles/oea-big-data-guide-1522052.pdf. 3. Zellkraftwerk, GmBH; Clinical Immunology; http://www.zellkraftwerk.com/files/Clinical_Immunology.pdf 4. PrattD, et.al.; NDEx, the Network Data Exchange; http://www.cell.com/cell-systems/abstract/S2405-4712(15)00147-7 5. CyREST: Turbocharging Cytoscape Access for External Tools via a RESTful API; http://f1000research.com/articles/4-478/v1

The National Resource for Network Biology is an NIH Biomedical Technology Research Center (BTRC)

supported by the National Institute of General Medical Sciences (NIGMS) under grant P41 GM103504.

Core Data

Classifiers

Normalizers

Intermediate Data

(Projectplasm)

Visualizations

ID Mapping

Methods in Experimental Description Language

Visualizations

Image Retrieval

Literature Searching

Zellkraftwerk

Z

Pathway Diagrams