drchriscole/godb_setup

Godb Setup

Attempt with Ubuntu 18.04

This is unique for M1 macs.

Followed the guide from https://mac.getutm.app/gallery/ubuntu-20-04, but importantly the 'Display' needs to be set to 'Console only' for the 18.04 iso from: http://old-releases.ubuntu.com/releases/bionic/ubuntu-18.04.4-server-arm64.iso

• once ubuntu is installed the 'Display' can be reset to 'Full Graphics'

• install a suitable desktop e.g. XFCE with

  sudo apt install tasksel sudo tasksel install xubuntu-desktop sudo reboot

Generic Instructions

Once you've got an Ubuntu 18.04 up and running continue here.

Package Installation

sudo apt-get install mongodb python-pymongo python-flask 
sudo apt-get install tabix libhts2 libhts-dev
sudo apt-get install python-pysam
sudo apt-get install python-pip       # for installing flask_wtf

Packages not available via apt-get

• flask_wtf

  • installed v0.14.3 due to a werkzueg dependency conflict with older versions

  • need to install from source via pip

    pip install -U https://github.com/wtforms/flask-wtf/archive/refs/tags/0.14.3.tar.gz
    

Config of GoDb codebase

This section will be obsolete as the codebase gets updated and is currently a working document.

• all the config files in /cfg need to be checked for valid paths
  • in particular 'common.cfg', 'godb.cfg' and 'webapp.cfg'
  • also affy.cfg, 'broad.cfg', 'exome.cfg' and 'illumina.cfg'
    • strikes me that they should be rationalised and be more agnostic
  • Now they all use the $GODBROOT environment variable instead

Mongo DB data loading

Firstly, mongod needs to be running. Currently this is on localhost:27017, but will need to be changed for production.

All the VCF data must be in this structure under $DATADIR (defined in cfg/common.cfg) which separates out each assay type:

$DATADIR/
  -> affy/
  -> broad/
  -> exome/
  -> illumina/

The vcf data should be bgzip compressed, tabix indexed and have two files per chromosome for all samples named:

chr1.vcf.gz     .. chr22.vcf.gz
chr1.vcf.gz.tbi .. chr22.vcf.gz.tbi

The data are loaded via three shell scripts which call other perl/python scripts:

bash load/sh/load_variants_from_vcf_files.sh $GODBROOT/cfg/broad.cfg
bash load/sh/load_samples.sh $GODBROOT/cfg/broad.cfg
bash load/sh/load_filepaths.sh $GODBROOT/cfg/broad.cfg

Repeat for each of the four assay types replacing the cfg file name to suit.

Load the webapp via bash webapp/sh/run_server.sh and the variants should be queryable from the interface on localhost:8085.

Job done!

Failed 20.04 Attempt

Followed this guide:

https://medium.com/@lizrice/linux-vms-on-an-m1-based-mac-with-vscode-and-utm-d73e7cb06133 https://mac.getutm.app/gallery/ubuntu-20-04

Start from Arm64 install of 20.04 Server

https://cdimage.ubuntu.com/releases/20.04/release/ubuntu-20.04.2-live-server-arm64.iso

Commands

After initial install of server

1. Install Xubuntu GUI

sudo apt install tasksel
sudo tasksel install Xubuntu-desktop
sudo reboot

2. Intall software packages

sudo apt-get install mongodb (v.3.6.8)
sudo apt-get install python (v2.7.18)

End no python-flask is available for python2 on 20.04. Dead end.