Sockeye FAQ

Sockeye FAQ

What platforms does Sockeye run on?

  1. Linux - We have tested on RedHat 7.2 and 8.0. We have also run Sockeye on SuSE 8.x Pro, but have tested it less thoroughly on this OS.
  2. Windows - We have tested on Win2K, and on XP Pro and Home.
  3. Mac OS X: When Java3D is available for OS X, we will make a version of Sockeye available for this OS. See http://developer.apple.com/java/faq/.

Which version should I use?

We highly recommend using the latest stable build or release version. The nightly builds might contain cutting edge new features, but did not pass our testing and QA process yet.

How do I install and start Sockeye?

To run Sockeye from Java WebStart, simply click on the link of the Sockeye version, Java WebStart will download all necessary files and start Sockeye automatically

What about Java?

  1. Sockeye is a Java application. For now, we only offer version that are bundled with a Java (1.4.1_01) and Java3D (1.3.1b1).
  2. We no longer offer a smaller download that has no bundled Java.
  3. You do not need to have Java or Java3D installed outside of Sockeye in order to run Sockeye. The installer will install the Java that Sockeye needs, and only Sockeye will use this Java.

How big are the files? How much disk space do I need?

  1. The download/installation sizes are -
    1. Linux - Download: ~50 MB. Installed application: ~75 MB.
    2. Windows - Download: ~21 MB. Installed application: ~50 MB.

Windows, Java3D, DirectX and OpenGL

  1. For Windows, we have offered installers with either DirectX or OpenGL versions of Java3D. Differences are discussed at http://www.j3d.org/implementation/java3d-OpenGLvsDirectX.html.
  2. At this point, we are not yet able to recommend one over the other. But we have tended to find that more systems accept the DirectX version, and more systems have problems with the OpenGL version. So, currently we offer only a DirectX Windows download.
  3. Sockeye may run better if you update your video card driver, if it is not current.
  4. The DirectX version requires a recent version of DirectX (see above j3d.org link). We have tested Sockeye with Microsoft's current version of DirectX, 9.0b, on Win2K and XP Pro. You can find the version installed on your computer by running Start > Run > dxdiag. You can get the latest DirectX version at http://www.microsoft.com/windows/directx/default.aspx.
  5. Note, however, that the DirectX version of Java3D does not support Antialiasing. This is likely only to matter for publication-quality, exported JPG images.
  6. In the past, we've solved some startup problems by switching the Display > Properties > Settings > Colour quality from '32 bit' to '16 bit'. Our sense is that this is less of a problem now, but if your 3D display does not appear cleanly and correctly, it's worth changing this setting to see if it helps.

How much RAM do I need?

  1. We have run Sockeye on PCs with between 192 MB and 1 GB of RAM.
  2. This release allows Sockeye's Java a maximum (heap) of 128 MB RAM, and we recommend having at least 256 MB installed.
  3. Working with large datasets may require more RAM than this.
  4. The online help discusses how you can change the maximum Java heap size setting by editing the LAX XML file in the install directory.
  5. To permit very large queries, Sockeye now uses 'on-demand' features, so you only load the annotation types that you ask for, and you can specify in the user_config.xml which types will be shown by default.
  6. To make it easier for you to work between very large and very small genomic ranges, we are adding 'semantic zooming' to Sockeye. This will automatically 'bin/unbin' feature types as you cross a feature-specific zoom threshold.
  7. When we are querying whole chromosomes we may well set the maximum Java heap to at least 512 KB, and will query in only a few features.

What kind of graphics card should I have?

  1. Because Sockeye is a 3D application, it demands much more powerful graphics hardware than a 2D genomic browser.
  2. The more 3D features you display, the more graphics power you will need to maintain a responsive 3D environment.
  3. The development team works with cards that are at least at the level of a GeForce2 MX. A card of this level still gives sluggish 3D zooms and rotations with a display showing, e.g., all 19.5K genes on the C. elegans genome.
  4. We have tested with cards from both ATI (Radeon) and nVidia. Text effects with Matrox cards can be somewhat mitigated by using 32-bit Z-buffering.

Is source code available?

Source code will be available for non-commercial users.

How do I...?

See 'How-tos' on the documentation page.

How do I query in really large genomic regions?

  1. It helps to have a fast CPU, a good graphics card, at least 512 MB RAM, and a fast Internet hookup.
  2. You will probably want to increase Sockeye's maximum Java heap size (e.g. to 512 MB) by editing the LAX XML file in the install directory. This is discussed in the online help.
  3. Watch the used and total Java memory in the right corner of Sockeye's status bar as you work, and compare its values to the maximum Java heap size setting.
  4. Use 'on-demand' features to query in only the annotation types that you need.
  5. In the near future, working with large regions will be easier. We are adding semantic zooming (SZ), which will automatically bin/unbin each type of feature as you cross a zoom threshold (you will be able to edit the threshold settings). While you will still appreciate a fast connection to an Ensembl server, this will mean -
    • more modest hardware will perform better
    • you should need to invest less in managing RAM settings
    • you should be able to move smoothly between displays showing large and small genomic regions. When you are zoomed far out, histograms should be a more meaningful display than individual features.

How do I query in Ensembl homologues for a gene?

  1. Query in an Ensembl data track with at least genes displayed.
  2. Right-click on a gene. The popup menu will tell you how many Ensembl homologues this gene has. If it has homologues, you will be able to follow the submenus to gene IDs to query in individual homologues. Alternatively, select 'Get all homologues' from the bottom of the first submenu.
  3. For each homologue you've requested, Sockeye queries in the annotations currently selected on the feature tree. The status bar will track the queries.
  4. When each homologue's annotations are returned, they will be shown on a new data track.

What are 'on-demand' features?

  1. When you query in Ensembl data, Sockeye only retrieves from the server the annotation types that you've selected in the 'feature tree'. Sockeye does not automatically query in all the Ensembl feature types that it is capable of.
  2. This minimizes query time, and helps you work with larger genomic regions with modest amounts of RAM.
  3. Note: While the most of on-demand features is functional, some details of how this interacts with 'navigation' operations are still being implemented.

Why the 'feature tree'?

  1. Together, the feature tree and the user_config.xml file, which defines the feature tree dynamically when you start Sockeye, let you -
    • organize feature types and subtypes in whatever hierarchical structure you want
    • control the how different annotation types are displayed in 3D
    • vary the level of detail in the control structure (tree) that you interact with.
  2. For example, they let you segregate groups of features; e.g. those specific to particular projects.
Page last modified Feb 06, 2007