Scribl: an HTML5 Canvas-based graphics library for visualizing genomic data over the web

Abstract

Motivation: High-throughput biological research requires simultaneous visualization as well as analysis of genomic data, e.g. read alignments, variant calls and genomic annotations. Traditionally, such integrative analysis required desktop applications operating on locally stored data. Many current terabyte-size datasets generated by large public consortia projects, however, are already only feasibly stored at specialist genome analysis centers. As even small laboratories can afford very large datasets, local storage and analysis are becoming increasingly limiting, and it is likely that most such datasets will soon be stored remotely, e.g. in the cloud. These developments will require web-based tools that enable users to access, analyze and view vast remotely stored data with a level of sophistication and interactivity that approximates desktop applications. As rapidly dropping cost enables researchers to collect data intended to answer questions in very specialized contexts, developers must also provide software libraries that empower users to implement customized data analyses and data views for their particular application. Such specialized, yet lightweight, applications would empower scientists to better answer specific biological questions than possible with general-purpose genome browsers currently available.

Results: Using recent advances in core web technologies (HTML5), we developed Scribl, a flexible genomic visualization library specifically targeting coordinate-based data such as genomic features, DNA sequence and genetic variants. Scribl simplifies the development of sophisticated web-based graphical tools that approach the dynamism and interactivity of desktop applications.

Availability and implementation: Software is freely available online at http://chmille4.github.com/Scribl/ and is implemented in JavaScript with all modern browsers supported.

Fig. 1.

The Rover genome browser built using the Scribl library. The three screenshots demonstrate the data views corresponding to the various semantic zoom levels. (A) High-level ‘birds-eye view’ of a large region of human chromosome 1, where sequencing read coverage from 1000 Genomes Project exome sequencing data is indicated by wiggle plots. (B) Data view at medium zoom level, showing annotation features and individual aligned reads as coordinate-type features. (C) Detailed data view at high resolution: the nucleotide sequence of each individual aligned read is shown. Additional information is revealed when hovered over. Hovering over a single-nucleotide polymorphism displays allele frequency and hovering over an aligned read displays insertions relative to the reference