The debt ceiling and budget kerfuffles seem to have delayed things in Bethesda somewhat, with the ripple effect of delaying our process of hiring someone to fill the JBrowse lead developer position, but we hope that will all be resolved soon (watch this space). In the meantime, check out the progress that Natalie has made with faceted browsing and hierarchical track organization — it is really exciting, and in fact seems quite obviously crucial, as genome databases naturally progress to having hundreds and even thousands of experimental annotation tracks.
The lead developer on the JBrowse team is moving on to pastures new: Mitch Skinner has accepted an offer to work at an exciting new medical genomics startup (which will, for the moment, remain nameless, but keep an eye out for their launch sometime in the next year).
Mitch hopes to continue contributing to JBrowse, but this also leaves an open position in the core team. Accordingly, we are now hiring.
New JBrowse release! This is version 1.2, which makes the server-side scripts much more scalable, includes a script for importing UCSC database dumps into JBrowse, and integrates Juan Aguilar’s touch code for the iPad/iPhone. Code is here:
These notes document changes since release 1.1 in September 2010.
(Release Notes for 1.1 are here.)
Most of the work in this release went into making JBrowse handle large
amounts of feature data better. Before, the amount of memory used
when processing BAM files was more than 10 times the size of the file;
now, the amount of memory required is fixed.
Other new features in this release:
Import of UCSC database dumps. A ucsc-to-json.pl script is now provided for taking database dumps from UCSC and creating a JBrowse instance using them. The “genePred” and “bed” track types are currently supported; “psl” tracks are not yet supported.
Touch. Juan Aguilar’s code for using JBrowse on an iOS device (iPhone, iPod touch, iPad) is now integrated. As of the current release, users wanting to use JBrowse on those devices have to navigate to a separate HTML page (touch.html) rather than the default index.html; i.e. the code does not currently detect touchscreen devices automatically.
Bug fixes. A number of bugs have also been fixed, including one that restricted the placement of the “data” directory, and a bug in wiggle rendering that caused spurious peaks or troughs at tile boundaries.
Known issues/limitations with this release:
Some additional CPAN modules are now required:
No JSON-level backward compatibility. If you are upgrading from an older version of JBrowse, you will have to regenerate all the JSON files on your server. This means wiping your jbrowse/data directory and re-running all server scripts (flatfile-to-json, biodb-to-json, wig-to-json, etc.) to regenerate your data from the original FASTA/GFF/BED/WIG files. We apologize for the inconvenience of this, but it is inevitable sometimes; we do aim to minimize the number of releases which are backwardly-incompatible in this way.
Something we get asked for a lot is data releases: tarballs of JBrowse JSON pre-generated from widely-used datasets, ready for you to plonk on your webserver, unzip, and serve to the world (perhaps after adding your own tracks, or making other modifications).
Data releases are good because they make it even easier to deploy JBrowse on your website, while illustrating what we consider best practices in using the JBrowse setup scripts.
Doing this right takes time, but it didn’t seem fair in the meantime to deprive people of the JSON files we’ve generated to run our demos. Particularly the human genome demo — there have been several requests for the data files underpinning that.
So, without further ado, here is a tarball of the UCSC hg19 annotation tracks. At least, some of them: we’ve tried to keep this tarball small. It’s 62 megabytes compressed; 330 megabytes uncompressed; it has the smallest 27 annotation tracks, and in order to keep the size down, we didn’t include actual sequence (just the locations of features like genes).
This demo should not be considered a final-quality data release. Some things aren’t set up as nicely as we’d like; for example, the outgoing links from features. However, if you’re interested in exploring using JBrowse for human applications (e.g. like genomesunzipped), this might help get you started.
(Update 2010/11/23: a bug in the tarball has been fixed; the link above now points to the fixed version)
(Update 2012/3/22: updated again to point to hg19mini on JBrowse 1.2.1)
The folks over at genomesunzipped.org are trying an interesting experiment in personal genomics. A group of experts in fields related to personal genomics are analysing their own genomes (via 23andMe) and making all the results available online. Naturally they chose JBrowse as their genome browser 🙂
It will be very interesting to see how this experiment in transparency plays out, and what kinds of discussion it provokes. More here.