This project is a complete rewrite of BioMAJ (http://biomaj.genouest.org).
BioMAJ (BIOlogie Mise A Jour) is a workflow engine dedicated to data
synchronization and processing. The Software automates the update cycle and the
supervision of the locally mirrored databank repository.
Common usages are to download remote databanks (Genbank for example) and apply
some transformations (blast indexing, emboss indexing,...). Any script can be
applied on downloaded data. When all treatments are successfully applied, bank
is put in "production" on a dedicated release directory.
With cron tasks, update tasks can be executed at regular interval, data are
downloaded again only if a change is detected.
More documentation is available in wiki page.
BioMAJ is python 2 and 3 compatible.
Edit global.properties file to match your settings. Minimal conf are database connection and directories.
biomaj-cli.py --config global.properties --status
biomaj-cli.py --config global.properties --bank alu --update
To migrate from previous BioMAJ 1.x, a script is available at:
https://github.com/genouest/biomaj-migrate Script will import old database to
the new database, and update configuration files to the modified format. Data directory is the same.
Migration for 3.0 to 3.1:
Biomaj 3.1 provides an optional micro service architecture, allowing to separate and distributute/scale biomaj components on one or many hosts. This implementation is optional but recommended for server installations. Monolithic installation can be kept for local computer installation.
To upgrade an existing 3.0 installation, as biomaj code has been split into multiple components, it is necessary to install/update biomaj python package but also biomaj-cli and biomaj-daemon packages. Then database must be upgraded manually (see Upgrading in documentation).
To execute database migration:
* Multiple remote protocols (ftp, sftp, http, local copy, ....)
* Data transfers integrity check
* Release versioning using a incremental approach
* Multi threading
* Data extraction (gzip, tar, bzip)
* Data tree directory normalisation
* Pre &Post processing :
* Advanced workflow description (D.A.G)
* Post-process indexation for various bioinformatics software (blast, srs,
fastacmd, readseq, etc…)
* Easy integration of personal scripts for bank post-processing automation
* Optional Administration web interface (biomaj-watcher)
* CLI management
* Mail alerts for the update cycle supervision
* Prometheus and Influxdb optional integration
* Optional consul supervision of processes
* Monolithic (local install) or microservice architecture (remote access to a BioMAJ server)
* Microservice installation allows per process scalability and supervision (number of process in charge of download, execution, etc.)
* Remote access:
* Optional FTP server providing authenticated or anonymous data access
* Debian: libcurl-dev, gcc
* CentOs: libcurl-devel, openldap-devel, gcc
Linux tools: tar, unzip, gunzip, bunzip
* mongodb (local or remote)
* elasticsearch (global property, use_elastic=1)
ElasticSearch indexing add advanced search features to biomaj to find bank
having files with specific format etc...
Configuration of ElasticSearch is not in the scope of BioMAJ documentation.
For a basic installation, one instance of ElasticSearch is enough (low volume of
data), in such a case, the ElasticSearch configuration file should be modified
node.name: "biomaj" (or any other name)
After dependencies installation, go in BioMAJ source directory:
python setup.py install
pip install biomaj biomaj-cli biomaj-daemon
You should consider using a Python virtual environment (virtualenv) to install BioMAJ.
In tools/examples, copy the global.properties and update it to match your local
The tools/process contains example process files (python and shell).
You can use BioMAJ with Docker (genouest/biomaj)
docker pull genouest/biomaj
docker pull mongo
docker run --name biomaj-mongodb -d mongo
# Wait ~10 seconds for mongo to initialize
# Create a local directory where databases will be permanently stored
docker run --rm -v local_path:/var/lib/biomaj --link biomaj-mongodb:biomaj-mongodb osallou/biomaj-docker --help
Copy your bank properties in directory *local_path*/conf and post-processes (if any) in *local_path*/process
You can override global.properties in /etc/biomaj/global.properties (-v xx/global.properties:/etc/biomaj/global.properties)
No default bank property file or process are available in the container.
Examples are available at https://github.com/genouest/biomaj-data
Execute unit tests
Execute unit tests but disable ones needing network access
nosetests -a '!network'
InfluxDB can be used to monitor biomaj. Following series are available:
* biomaj.banks.quantity (number of banks)
* biomaj.production.size.total (size of all production directories)
* biomaj.workflow.duration (workflow duration)
* biomaj.production.size.latest (size of latest update)
* biomaj.bank.update.downloaded_files (number of downloaded files)
* biomaj.bank.update.new (track updates)
Biomaj uses libcurl, for sftp libcurl must be compiled with sftp support
To delete elasticsearch index:
curl -XDELETE 'http://localhost:9200/biomaj_test/'
Special thanks for tuco at Pasteur Institute for the intensive testing and new
Thanks to the old BioMAJ team for the work they have done.
BioMAJ is developped at IRISA research institute.
## Needs database upgrade
If using biomaj-watcher, must use version >= 3.1.0
Feature #67,#66,#61 switch to micro service architecture. Still works in local monolithic install
Fix some configuration parameter loading when not defined in config
Fix HTTP parsing parameters loading
Fix download_or_copy to copy files in last production release if available instead of downloading files again
Manage user migration for micro services
Feature #74 add influxdb statistics
Feature #65 add a release info file at the root of the bank which can be used by other services to know the latest release available
Feature #25 experimental support of rsync protocol
Add rate limiting for download with micro services
Limit email size to 2Mb, log file may be truncated
Fix #55: Added support for https and directhttps
Add possibility to define files to download from a local file with remote.list parameter
Fix visibility modification (bug deleted the bank properties field)
Fix #65 Add release file in bank dir after update
Add md5 or sha256 checksum checks if files are downloaded and available
Fix missing README.md in package
Fix #53 avoid duplicates in pending databases
Add migration method to update schema when needed
Manage HTTP month format to support text format (Jan, Feb, ...) and int format (01, 02, ...)
New optional bank property http.parse.file.date.format to extract date in HTTP protocol following python date regexp format (http://www.tutorialspoint.com/python/time_strptime.htm)
Example: %d-%b-%Y %H:%M
Fix #47: save_as error with directhttp protocol
Fix #45: error with pending releases when release has dots in value
Do not use config values, trust database values #39
Fix #42: Add optional release.separator to name the bank directory bankname_release (underscore as default)
Fix #37: remote local files history from db and put it in cache.dir
Feature #38: add optional keep.old.sessions parameter to keep all sessions in database, even for removed releases
Feature #28: add optional release.format parameter to specify the date format of a release
Fix in method set_owner
Force release to be a str
Fix #32: fix --from-task issue when calling a meta process
Fix #34: remove release from pending when doing cleanup of old sessions
Remove logs on some operations
Add --status-ko option to list bank in error state
Fix #36 manage workflows over by error or unfinished
Fix #27: Thread lock issue during download
New optional attribute in bank properties: timeout.download
HTTP protocol fix (deepcopy error)
Fix index deletion on bank removal
Fix lock errors on dir creation for multi-threads,
pre-create directroy structure in offline directory
Fix #26: save error when too many files in bank
Fix in session management with pre and rm processes
Fix #23: Check workflow step name passed to
Fix #24: deprecated delete_by_query method in elasticsearch
Add some controls on base directories
Change dir to process.dir to find processes in subdirs
If all files found in offline dir, continue workflow with no download
Remove extra log files for bank dependencies (computed banks)
Fix computed bank update when sub banks are not updated
Fix #15 when remote reverts to a previous release
Feature #16: get possibility not to download files (for computed banks for
example). Set protocol='none' in bank properties.
Fix on --check with some protocols
Fix #21 release.file not supported for directhttp protocol
Feature #22: add localrelease and remoterelease bank properties to use the
remote release as an expression in other properties
=> remote.dir = xx/yy/%(remoterelease)s/zz
Feature #17,#20: detect remote modifications even if release is the same
new parameter release.control (true, false) to force a check
even if remote release (file controlled or date) is the same.
Fix on 'multi' protocol
Fix on "save_as" regexp when remote.files starts with a ^ character.
Fix thread synchro issue:
during download some download threads could be alive while main thread continues worflow
the fix prevents using Ctrl-C during download
if subtask of workflow fails, fail main task
osallou authored 14 hours ago
do not test index if elasticsearch is not up
add http proxy support
retry uncompress once in case of failure (#13)
Reindent code, pep8 fixes
Various fixes on var names and OrderedDict suport for Python < 2.7
Merge config files to be able to reference global.properties variables in bank
property file in format %(xx)s
Use ConfigParser instead of SafeConfigParser that will be deprecated
Add option --remove-pending to remove all pending sessions and directories
Add process env variables logdir and logfile
Fix Unicode issue with old versions of PyCurl.
Fix removal workflow during an update workflow, removedrelease was current
Fix shebang of biomaj-cli, and python 2/3 compat issue
Update code to make it Python 3 compatible
Use ldap3 library (pure Python and p2,3 compatible) instead of python-ldap
get possiblity to save downloaded files for ftp and http without keeping full
remote.files can include groups to save file without directory structure,
or partial directories only, examples:
remote.files = genomes/fasta/.*\.gz => save files in offline directory, keeping remote structure offlinedir/genomes/fasta/
remote.files = genomes/fasta/(.*\.gz) => save files in offline directory offlinedir/
remote.files = genomes/(fasta)/(.*\.gz) => save files in offline directory offlinedir/fasta
TODO: Brief introduction on what you do with files - including link to relevant help section.