This is for the ultra-impatient and assumes:
- have conda already, if not read Step 1 and 2.
- have a licence to FontAwesome-Pro, if not read Step 3
- dont care about VENUS or gene name route, if not read Step 4
- you want to use sqlite, not postgres. if not read Step 5
- dont care about static downloads, if not read step 6
Install modules in a folder called michelanglo:
mkdir michelanglo
cd michelanglo/
git clone --recursive https://github.com/matteoferla/MichelaNGLo.git app
git clone https://github.com/matteoferla/MichelaNGLo-protein-analysis analysis
git clone https://github.com/matteoferla/MichelaNGLo-transpiler transpiler
cd analysis
python3 setup.py install
#python3 create.py & #do this to get protein data. else:
mkdir ../analysis
mkdir ../analysis/reference
touch ../analysis/reference/pdb_chain_uniprot.tsv
# end of hacky way round.
cd ../transpiler
python3 setup.py install
cd ../app
python3 setup.py install
npm i puppeteer # installs puppeteer, but you will have to decide about the chromium sandboxing
npm i fs-extra
## For Jquery to be added to the offline download it needs to be compiled.
CURRENT_PATH=`pwd`
cd app/michelanglo_app/static/ThirdParty/jquery
npm run build
cd $CURRENT_PATH
# do not want sqlite? change the env variable accordingly and read below
cp demo.db mike.db
PROTEIN_DATA='../protein-data' SECRETCODE='needed-for-remote-reset' SQL_URL='sqlite:///mike.db' SLACK_WEBHOOK='https://hooks.slack.com/services/xxx/xxx/xxx' python3 app.py > ../mike.log 2>&1
The last command runs it (to run in dev mode add --d).
The admin user in the demo.db file is admin with password admin.
Additional adding the env var SENTRY_DNS_MICHELANGLO='https://[email protected]/xxx' will send errors.
The Slack webhook isn't optional, but giving a dummy value will just make errors, but not crash it.
This is why there are many commands to copy.
** Do not jump the gun and git clone recursively** this repository without altering FontAwesome Pro requirement if needed (see below).
Michelanglo has lots of moving parts...
- Python 3.7
- Pyramid
- Waitress (Apache: reverse proxy)
- PyMOL (see note)
- Biopython
- APScheduler
- Postgres
- external purpose written modules:
- NodeJS
- Puppeeter
Python 3.7: This repo uses f-strings and the author is unwilling to make it backwards compatible. DB: Postgres is more robust and secure than SQLite. But if you are just running this for yourself, be lazy & go SQLite.
- Mako-templated HTML
- NGL
- JQuery
- Boostrap 4.3
- FontAwesome Pro
- Bootstrap Tour (mod)
FontAwesome: I have a pro licence, but a free version is available. So a wee change is required to .gitmodules (see below).
Linux packages. apt for Ubuntu, yum for CentOS.
sudo apt install nodejs
sudo apt install npm
And either...
sudo apt install sqlite
or...
sudo apt install postgres
Mac packages
brew install nodejs
brew install npm
brew install wget
either...
brew install sqlite
or...
brew install postgres
Installing postgres on a networked windows machine is a satanic endevour.
Consider a Docker container with Alpine, a VM with Ubuntu, Rasperry pi with Raspian (use Berryconda) or an old smartphone with AfterMarketOS.
All the documentation here works on Ubuntu, CentOS and MacOS and everything in Windows it is a bit trickier. But the excecutables will have .exe suffixes and are in Scripts folder C:\Users\yournamehere\AppData\Local\Continuum\anaconda3\Scripts\pip3.exe say for your regular install, your virtual env will be wherever you put it.
This app requires Python3 compiled PyMOL. The best option is using Conda. Otherwise it needs to be compiled (instructions). So the best bet is to install anaconda3:
wget https://repo.anaconda.com/archive/Anaconda3-2019.10-MacOSX-x86_64.sh
bash Anaconda3-2019.10-MacOSX-x86_64.sh -b
conda init
conda config --set auto_activate_base true
conda update conda
If you want to make env —your call
conda create -n env python=3.7 anaconda
conda activate env
Install what you need:
conda install -c schrodinger pymol
conda install -c conda-forge -y biopython
#conda install -c conda-forge -y rdkit
For fetching proteins, michelanglo_app requires https://github.com/matteoferla/MichelaNGLo-protein-module.
Both the protein module and Michelanglo require a PyMOL manipulation script, whcih is separate as the protein parsing module works without Michelanglo. https://github.com/matteoferla/MichelaNGLo-transpiler for more.
mkdir michelanglo
cd michelanglo/
git clone https://github.com/matteoferla/MichelaNGLo-protein-analysis.git protein-analysis
cd protein-module
python3 setup.py install
git clone https://github.com/matteoferla/MichelaNGLo-transpiler transpiler
cd ../transpiler
python3 setup.py install
Do you have FontAwesome Pro?
git clone --recursive https://github.com/matteoferla/MichelaNGLo.git app
cd ../app
python3 setup.py install
Installing it isn't needed for normal operations, just esthetics.
Else:
git clone https://github.com/matteoferla/MichelaNGLo.git app
# https://github.com/FortAwesome/Font-Awesome
sed -i 's/Font-Awesome-Pro\.git/Font-Awesome\.git/g' .gitmodules
git submodule update --init --recursive
python3 setup.py install
Note that you'll also need to change all instances of the class far with fas in the templates
by adding <script>$('.far').addClass('fas').removeClass('far')</script> near the bottom of templates/layout_components/layout.mako
(There is a common giving more info within there).
The option of searching Font-Awesome-Pro or FA5 in GitHub is not recommended
as these public forks are in violation of copyright of a small startup team not a megacorporation.
It also uses a protein module to allow gene name querying.
This module uses a lot of data. That unfortunately I cannot keep as a repo for you to download. However. This step is optional: if not done, gene retrieval will not work.
This module has lots of cool stuff. I might be worth your while checking it out.
see https://github.com/matteoferla/MichelaNGLo-protein-module
Also, if you plan to mod Michelanglo do not clone the protein module in Michelanglo or your IDE will go extremely slow.
To get everything...
cd ../protein-module
python3 create.py
This will save all the data it will download and parse to this folder. This will take a whole day. For licencing issues, Phosphosite plus data needs to be downloaded manually.
However, if there is a species you are interested in, email me and I can save you the bother.
The database needs starting, for SQLite make a copy of demo.db or do the following:
touch mike.db
SQL_URL=sqlite:///mike.db alembic -c development.ini revision --autogenerate -m "init"
SQL_URL=sqlite:///mike.db alembic -c development.ini upgrade head
If using postgres the environment variable needs to be SQL_URL=postgresql://name_of_owner_user_you_made_for_the_db_that_is_not_postgres:its_password@localhost:5432/name_of_db
Obviously, nothing ever goes smoothly. If you get an error with the second line (the upgrade) edit the file michelanglo_app/alembic/versions/xxxx.py if you get:
- an error about explicit contraint names: change all
sa.Boolean()tosa.Boolean(create_constraint=False). SQLite does not know about Booleans. - ... ?
For a more robust system, use postgres (as used in our version).
In order to get thumbnails of the protein in the galleries, or for when you share your protein on Twitter or Facebook, nodejs with puppeteer is required.
npm install puppeteer
npm install fs-extra
Also, some of the submodules in michelanglo_app/static/ThirdParty need building —JQuery.
But this is only required for static offline downloads.
Having an config file with sensitive data on GitHub is a no-no, so there are two options. The first is copying the config file and amend it.
cp production.ini actual.ini
python app.py --config actual.ini > ../mike.log 2>&1
The other is to use environment variables, which will override the config. To pass an environment variable to a command one can do
FOO=foo command arg1 arg2;
The variable $FOO will be available within the command run, but not outside.
There are lots of env variables usable and for some reason I decided to obey conventions, so here is the mapping:
| config name | environ name |
|---|---|
| michelanglo.protein_data_folder | MICHELANGLO_PROTEIN_DATA |
| michelanglo.user_data_folder | MICHELANGLO_USER_DATA |
| michelanglo.secretcode | MICHELANGLO_SECRETCODE |
| sqlalchemy.url | MICHELANGLO_SQL_URL |
| sentry.data_source_name | MICHELANGLO_SENTRY_DNS |
| puppeteer.executablePath | PUPPETEER_CHROME |
| slack.webhook | SLACK_WEBHOOK |
| michelanglo.admin_email | MICHELANGLO_ADMIN_EMAIL |
| michelanglo.server_email | MICHELANGLO_SERVER_EMAIL |
These are:
Where did you put the protein
PROTEIN_DATA='/home/apps/protein-data'
code is used to give the command to reset mike:
SECRETCODE='1234567890'
DB URL
SQL_URL='postgresql://username:password@localhost:5432/app_users'
(opt) Sentry
SENTRY_DNS_MICHELANGLO='https://xxxxx'
Slack webhook to keep you in the loop. Note that to get a slack webhook you don't go in your normal page, but in api.slack.com
SLACK_WEBHOOK='xxxxxxxxxxx'
So a bash variable is declared without spaces a="hello world" and then you can call it echo $a. These will not be available outside of the current session, unless you export $a.
Alternative you can run the application you want to feed the env variable without leaving a trace(ish) by a="hello world" python myscript
SQL_URL=xxx;SECRETCODE=xxx;SLACK_WEBHOOK=xxx;PROTEIN_DATA=xxx python3 app.py --d
Run the script and make a user called admin. This is a one time only option.
The users trashcan gets generated automatically when a guest makes a view and is blacklisted along with guest and Anonymous.
Set up a system daemon (as done in the SGC server), or a cron job to make sure it comes back upon system failure. Also, the app.py serves on port 8088.
For Venus is required. PyRosetta is for academic use only and the username and password is obtained from https://www.rosettacommons.org/software/license-and-download
curl -u $PYROSETTAUSER:$PYROSETTAPASSWORD https://graylab.jhu.edu/download/PyRosetta4/archive/release/PyRosetta4.Release.python3.8.ubuntu/PyRosetta4.Release.python3.8.ubuntu.release-295.tar.bz2 -o /content/a.tar.bz2
tar -xf a.tar.bz2
pip3 install -e PyRosetta4.Release.python3.8.ubuntu.release-295/setup/
rm a.tar.bz2