The UCSC Genome Browser is capable of displaying both the BAM and CRAM file formats. While BAM files contain all sequence data within a file, CRAM files are smaller by taking advantage of an additional external "reference sequence" file. This file is needed to both compress and decompress the read information.

Since CRAM files are more dense than BAM files, many groups are switching to the CRAM format to save disk space. For CRAM tracks to load there is an expectation that the md5sum of the reference sequence used to create the CRAM will be in the CRAM header. A file with a matching md5sum is also expected to be accessible from the EBI CRAM Reference Registry.

Please also note that just as a BAM file requires an associated BAM.bai index file, a CRAM file will require an associated CRAM.crai index file in the same location to load.

Example One

Here is an example CRAM track that displays around the gene SOD1 on hg19 that can be cut and pasted ast text into the Custom Tracks page:

track type=bam db=hg19 name=exampleCRAM bigDataUrl=http://genome.ucsc.edu/goldenPath/help/examples/cramExample.cram

Clicking this following link will also load the above track. The information following hgct_customText is equivalent to pasting the text in to the Custom Tracks page:

http://genome.ucsc.edu/cgi-bin/hgTracks?db=hg19&position=chr21%3A33031597-33041570&hgct_customText=track%20type=bam%20name=exampleCRAM%20bigDataUrl=http://genome.ucsc.edu/goldenPath/help/examples/cramExample.cram

Since the loading of CRAM data requires the specific reference sequence used to create the CRAM file, it is very important that the exact same reference sequence is used for compression and decompression. When a CRAM file is first loaded on a given chromosome, a check for the preexistence in a special browser "cramCache" directory of the specified reference md5sum will take place. If the reference sequence information specific for that CRAM for the currently viewed chromosome region does not exist, a message will display about the file not being found along with a note about downloading the reference from the EBI CRAM Reference Registry if it is available. A refresh of the page once the download is complete will display the CRAM data as if it were a BAM file.

Example Two

Clicking the following image will load a CRAM file from the 1000 Genomes Project.

This CRAM display takes advantage of using the new "density graph" feature where the bam.cram reads are displayed as a bar graph by checking the box next to "Display data as a density graph" on the Custom Track Settings page.

Example Three

The CRAM format is also supported in track hubs. Below is an example trackDb.txt stanza that would display a CRAM files from the 1000 Genomes Project. To learn more about using Track Hubs see the User Guide and associated Quick Start Guides to building hubs. Note that type bam is used to display CRAM files in hubs, just as type bam is used in custom CRAM tracks.

track cram61
type bam
shortLabel HG00361
longLabel This CRAM file is from the 1000 Genomes Project HG00361
visibility pack
bigDataUrl ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/phase3/data/HG00361/exome_alignment/HG00361.mapped.ILLUMINA.bwa.FIN.exome.20120522.bam.cram

References

• CRAM toolkit
• CRAM reference registry
• CRAM format specification (version 3.0)
• Using CRAM within Samtools
• International Genome Sample Resource (IGSR) CRAM Tutorial
• Dave Tang blog post about "BAM to CRAM"

Sharing Your Data with Others

If you would like to share your CRAM data track with a colleague, learn how to create a URL by looking at Example 11 on this page.

Activating CRAM support for the Genome Browser

To find documentation on how to set up CRAM support on a mirror of the UCSC Genome Browser please see this following README.cram file.