If you do not have access to git, you can obtain our latest API code as a gzipped tarball:
Note: the API version needs to be the same as the databases you are accessing, so please use git to obtain a previous version if querying older databases.
Entire databases can be downloaded from our FTP site in a variety of formats. Please be aware that some of these files can run to many gigabytes of data.
To facilitate storage and download all databases are GNU Zip (gzip, *.gz) compressed.
About the data
The following types of data dumps are available on the FTP site.
- FASTA sequence databases of Ensembl gene, transcript and protein
model predictions. Since the
FASTA format does not permit sequence annotation,
these database files are mainly intended for use with local sequence
similarity search algorithms. Each directory has a README file with a
detailed description of the header line format and the file naming
- Masked and unmasked genome sequences associated with the assembly (contigs, chromosomes etc.).
- The header line in an FASTA dump files containing DNA sequence consists of the following attributes : coord_system:version:name:start:end:strand This coordinate-system string is used in the Ensembl API to retrieve slices with the SliceAdaptor.
- Coding sequences for Ensembl or ab initio predicted genes.
- cDNA sequences for Ensembl or ab initio predicted genes.
- Protein sequences for Ensembl or ab initio predicted genes.
- Non-coding RNA gene predictions.
- Annotated sequence
- Flat files allow more extensive sequence annotation by means of feature tables and contain thus the genome sequence as annotated by the automated Ensembl genome annotation pipeline. Each nucleotide sequence record in a flat file represents a 1Mb slice of the genome sequence. Flat files are broken into chunks of 1000 sequence records for easier downloading.
- All Ensembl MySQL databases are available in text format as are the SQL table definition files. These can be imported into any SQL database for a local installation of a mirror site. Generally, the FTP directory tree contains one directory per database. For more information about these databases and their Application Programming Interfaces (or APIs) see the API section.
- Gene sets for each species. These files include annotations of both coding and non-coding genes. This file format is described here.
- GFF3 provides access to all annotated transcripts which make up an Ensembl gene set. This file format is described here. dt class="bg1">GVF (variation data)
- GVF (Genome Variation Format) is a simple tab-delimited format derived from GFF3 for variation positions across the genome. There are GVF files for different types of variation data (e.g. somatic variants, structural variants etc). For more information see the "README" files in the GVF directory.
- VCF (variation data)
- VCF (Variant Call Format) is a text file format containing meta-information lines, a header line, and then data lines each containing information about a position in the genome. This file format can also contain genotype information on samples for each position. More details about the format and its specifications are available here.
- VEP (variation data)
- Compressed text files (called "cache files") used by the Variant Effect Predictor tool. More information about these files is available here.
- BED format files (comparative data)
Constrained elements calculated using GERP are available in BED format. For more information see the accompanying README file.
BED format is a simple line-based format. The first 3 mandatory columns are:
- chromosome name (may start with 'chr' for compliance with UCSC)
- start position. This is a 0-based position
- end position.
The entire Ensembl API is gzipped and concatenated into a single TAR file. This is updated daily.