You can run the GLUE command line tool on CLIMB infrastructure. Benefits:
There is a special CLIMB GLUE VM glue.covid19.climb.ac.uk for running GLUE. You can ssh to it using the same user ID (e.g. climb-covid19-singerj) and private key file (e.g. climb_cog_id_rsa) you would use for COG-UK CLIMB generally:
If this does not work you should post on the COG-UK Slack #account-requests channel.
$ ssh -i climb_cog_id_rsa climb-covid19-singerj@glue.covid19.climb.ac.uk
Once you are able to access the GLUE VM you need to contact one of the GLUE VM admininstrators and have them set up your GLUE database for you. They will send you a new username and password for this database.
You then need to connect to the GLUE VM and perform a one-time task to set up your personal GLUE instance within the VM:
$ initGlue.sh -u <glueDbUsername>-p <glueDbPassword>
(note that glueDbUsername and glueDbPassword here are credentials that the GLUE VM administrator will have sent you)
You should see some log messages ending with:
GLUE installation complete
Log in again or run source ~/.bash_profile to add gluetools.sh to your PATH
Now do:
$ source ~/.bash_profile
This adds the gluetools.sh command to your PATH. On future logins to the VM it will always be there.
It is possible for one user of the CLIMB VM to run multiple isolated GLUE installations each with their own separate databases. This could be useful if for example on the HOCI project you are responsible for multiple HOCI sites. If you require this set-up, contact one of the GLUE VM administrators on the COG-UK Slack #cov-glue channel. Do not register for multiple COG-UK/Majora accounts.
You now have a personal GLUE instance but it contains no data. You have two options:
$ installGlueProject.sh cov_glue
installs the dataset backing the CoV-GLUE web application, based purely on GISAID data.$ installGlueProject.sh coguk_cov_glue
installs a larger dataset, containing cov_glue plus COG-UK sequences/metadata and various items relating to the HOCI project.The cov_glue dataset is being updated approximately weekly. The coguk_cov_glue dataset is no longer being updated.
To give a flavour of what CoV-GLUE can do, try this:
$ gluetools.sh
GLUE> project cov
OK
GLUE> list sequence -w "m49_country.id = 'BEL' and collection_date < #gluedate(01-Mar-2020)" sequenceID collection_date cov_glue_lineage
OK
GLUE> module covFastaProteinAlignmentExporterSeqIdOnly export AL_GISAID_CONSTRAINED -r REF_MASTER_WUHAN_HU_1 -f NSP13 -l 10 20 -w "sequence.m49_country.m49_region = 'asia' and sequence.cov_glue_lineage like 'A%'" -p
...
GLUE> quit
The list sequence command listed the GISAID ID, collection date and lineage of all sequences collected in Belgium before 1st March.
The longer command exported a protein alignment of nsp13, codons 10 to 20, for all lineage A sequences from Asia.
Many other things are possible. Anything where you are querying some aspect of a multiple sequence alignment together with metadata is within the scope of CoV-GLUE.