GSoC 2014 proposal version 2

index.md

Overview

The goal of my project is to build interactive visualization, and still allows fast prototyping, customizability, and integration to other SciRuby components. I think we can complete this goal with the client-server model shown below: server provides JSON data throgh REST API, and client get it with Ajax and generate plots based on it. Server-side gem will be build with SciRuby components like NMatrix, and client-side will use D3.js for plotting. This model is inspired by Prof. Karl Broman's work. (http://www.biostat.wisc.edu/~kbroman/D3/)

Project plan

Client side

Base gem

• This gem is tentatively named SciPlot.
• SciPlot provides low level API for plotting like 'line,' 'rect,' and 'dot.'
• This can generate plots that imports data from Ajax data source. D3.js already has asynchronous loading API like d3.json, and we can add this feature easier.

Plotting gems

• We will create 3 new gems for plotting - Plotting for general use, Biology, and 3D plotting.
• 3D plotting library will not depend on SciPlot because it do not use 2D API. I will implement it only for practice of Ajax loading and implementation of Ruby wrapper for JavaScript.

Supported plots

• 10 charts for SciPlot STL (general use charts)
• Scatter
• Bar
• Box
• Line
• and 6 more
• BioPlot
• HeatMap
• Network
• Matrix
• and more
• 3D Plots
• see here

Mock code

require 'sciplot'
scatter = SciPlot::Scatter.new([1,2,3])
scatter.show() #-> show charts on IRuby notebook

require 'bioplot'
genomemap = BioPlot::GenomeMap.new(some_data) # internally using SciPlot's DSL
genomemap.show() #-> IRuby

Server side

Overview

• SciRuby-Server consists of 4 parts, Server, REST API, Web interface and DataBase.
• We can deploy it into www server, or publish to closed network, or use in local environment.
• SciRuby::Dataframe internally use REST API and push or pull data through it.
• Users need not to be aware of where they store their data.
• I will implement it with Sinatra and non-SQL Database. I will also use Thin for testing server in local environment.

Mock code

require 'sciruby'
require 'sciplot'
db = SciRuby::DataFrame.new(column = some_data, ...)
db.connect(url) # do configuration when connecting to SciRuby-Server on Web.
SciPlot::Scatter.new(db) # generates plots from web source.

Situations

Local environment

Public environment

• After publishing SciRuby-Server, you can also update data through REST API, using Dataframe of other wrapper.
• REST-API also enables us to automatically update data. (like sensor and another server)
• Some plots generated with our Client-side gem dinamically import data from SciRuby-Server.

Roadmap

• implementation of client-side libraries (2-3 weeks)
  • First I will implement general plots for SciPlot STL, and implement reusable API based on that work.
• implementation of wrapper gems for JS library (at the same time as previous step)
• implementation of more advanced library ( 1-2 weeks)
• BioPlot and 3D plot.
• implementation of SciRuby-Server and SciRuby::Dataframe (2 weeks)
• experimental implementation of SciRuby DB based on Dataframe (1 week)
  • Thin + Sinatra App + some DB