Metadome Web server
Here you may find all details on how to set up the MetaDome web server in your own environment.
The MetaDome webserver has only been tested for running in a Linux-based OS (tested on Ubuntu 16.04, Ubuntu 18.04 and Centos 7), but as it is fully containerized in docker it should in theory be possible to run it on any other OS. Please ensure you have the following software installed on your machine:
The MetaDome web server makes use of the following data resources
MetaDome uses version v19 of Gencode for GRCH37. Available here, please download the following files:
gencode.v19.annotation.gtf gencode.v19.annotation.gff3 gencode.v19.pc_transcripts.fa gencode.v19.pc_translations.fa gencode.v19.metadata.SwissProt ucsc.gencode.v19.wgEncodeGencodeBasic.txt
Please put these in your preferred data storage folder under '/Gencode/'
MetaDome was tested with UniProt version 2016_09. Available here, please download the following files:
uniprot_sprot.dat.gz uniprot_sprot_varsplic.fasta.gz uniprot_sprot.fasta.gz
Unzip 'uniprot_sprot_varsplic.fasta.gz' and 'uniprot_sprot.fasta.gz' and combine them to a new file: uniprot_sprot_canonical_and_varsplic.fasta
Next you should build a blast database from this new file named 'uniprot_sprot_canonical_and_varsplic.fasta'. See here for a further explanation on how to do this if you are unfamiliar with that.
Next, put all of this in your preferred data storage folder under '/UniProt/'
MetaDome was tested with version 20180603 for GRCh37, but later versions may work as well as long as the GRCh37 remains unchanged. From here, please download the following files: clinvar.vcf.gz clinvar.vcf.gz.tbi
And put this in your preferred data storage folder under '/ClinVar/GRCh37/'
MetaDome was tested with version r2.0.2 for GRCh37, but later versions may work as well as long as the GRCh37 remains unchanged. From here, please download the following files:
And put this in your preferred data storage folder under '/gnomAD/'
MetaDome was tested with PFAM version 30.0. From here, please download the following files:
Unzip 'Pfam-A.hmm.gz' to Pfam-A.hmm. We used a Python script (available upon request) to seperate each of the PFAM alignments contained within 'Pfam-A.hmm.dat.gz' to a folder : 'alignment/PFXXXXX/full' where XXXXX corresponds to the PFAM identifier for faster querying and the alignment is contained in the file 'full'. MetaDome webserver expects the files to be organised as such.
Put all this in your preferred data storage folder under '/PFAM/'
Clone the repository and go into the project folder:
git clone https://github.com/cmbi/metadome.git cd metadome
First configure the volumes (at the <ABSOLUTE PATH>) to correspond with your data folders in the docker-compose.yml (lines 72, 78, 84, 90, 96, 102).
Also make sure to create a folder in your data folder specifically to contain the MetaDome mapping database and at the path to line 108 in the docker-compose.yml. This database is to be created on a first run (see First time set-up below).
(Optional) Credentials configuration
If you are planning to expose the MetaDome server to a public adress, please make sure you get the security in order.
- Change the variable `POSTGRES_PASSWORD` for the postgresql database in `./metadome/postgres_credentials.py` - Change the variable `SECRET_KEY_CRED` in `./metadome/flask_app_credentials.py`
Otherwise you will be using default passwords and API secrets.
Running the server
First time set-up
Run the following command to execute the install script:
docker-compose run app python install.py
Note: Launching the server this way, by creating the database from scratch, will first make it generate all mappings between gencode, swissprot and Pfam. This process, depending on your configuration, may take 2 weeks. If you require a pre-build database, do not hesitate to contact us.
Starting the server
If you followed the above steps, you can now run the webserver via the command:
docker-compose docker-compose.yml up -d
And tear it down via:
docker-compose -f docker-compose.yml stop
Citing MetaDome server
Please cite the MetaDome web server article :
MetaDome: Pathogenicity analysis of genetic variants through aggregation of homologous human protein domains. Laurens Wiel, Coos Baakman, Daan Gilissen, Joris A. Veltman, Gert Vriend and Christian Gilissen. Human Mutation. (2019) 1-9, 10.1002/humu.23798
If you want to provide feedback please have a look at our existing issues (and if necessary, create a new issue).