Computational protocol: Software tools for visualizing Hi-C data

Similar protocols

Protocol publication

[…] All five tools differ fairly substantially in their functionality but also in how they are implemented. In particular, although all of the tools are freely available, only Hi-Browse, the Epigenome Browser, and Juicebox are open source. Furthermore, the Epigenome Browser and Juicebox can be installed to run on the user’s local computer, which circumvents the need to access online servers through the internet. This is desirable for analyses that require confidentiality or significant computational resources. Local installation for Juicebox requires only a 64-bit Java distribution, whereas installation of the Epigenome Browser depends on multiple software packages and server services, described in detailed, step-by-step instructions in the corresponding manual.All of the tools provide a graphical user interface that is available through a web browser interface or via Java Web Start, and thus requires no or minimal installation. Unless a local installation is performed, all tools also require an internet connection. Access to tools that use a web browser interface is available through any operating system. For local installations, the Epigenome browser supports Linux and MacOS operating systems.Documentation is provided for each of the five tools, although documentation of the 3D Genome Browser is being updated at present. The Epigenome Browser has its own wiki page that explains how to create and manage files for storing track information. Juicebox and the Epigenome browser have active online discussion groups that are maintained by the tool developers.For each visualization tool, we profiled the speed of two important operations: loading user data and visualizing loci of sizes that are appropriate for both browser-based and heat-map-based tools (Table ). Many factors, such as internet connection speed and server load, make it challenging to set up an exact benchmarking protocol; thus, we only report the approximate speed of loading operations, on the order of seconds, minutes or hours, and we report an average duration for visualization tasks. For benchmarking, we set the resolution parameter to either 40 kb or 50 kb, commonly used resolutions that strike a balance between sparsity and detail. We found that Juicebox, the Epigenome Browser and the 3D Genome Browser process user data in binary formats in a few seconds. Hi-Browse and my5C do not support loading of a complete dataset at these resolutions, instead the user must upload the Hi-C contact matrix corresponding to the region of interest. The average times required to visualize 1 Mb and 10 Mb heat maps showed that tools that do not use a browser framework are faster, with Juicebox and my5C the fastest tools. Browser-based tools are generally slower, especially for 10 Mb loci, consistent with the browser-based tools’ intended focus on local visualizations. We stress that user experience might differ from our benchmark due to differences in data sets, internet bandwidth and other parameters; thus, we offer this benchmark as a general guideline rather than an absolute measure of speed. […]

Pipeline specifications