The purpose of this repository will be to help track designs and make Rosetta job submissions easier.
pandas
numpy
docopt
yaml
klab (python3_compatibility branch)
matplotlib version 3.1.3
pygobject (try conda install -c conda-forge pygobject)
pango (easiest to install via conda install -c anaconda pango)
pygtk (easiest to install via conda install -c vgauthier pygtk)
pygtk (conda install -c pkgw-roge gtk3)
seaborn
The easiest way to install all the dependencies is to use the included conda environment file.
(Mac) curl -O https://raw.githubusercontent.com/ckrivacic/roseasy/master/conda_env.yml
(Linux) wget https://raw.githubusercontent.com/ckrivacic/roseasy/master/conda_env.yml
conda env create --file conda_env.yml
PyRosetta is not included, so make sure that is installed in the same environment.
To install RosEasy:
git clone https://github.com/ckrivacic/roseasy.git
python setup.py develop
roseasy <command> <arguents> [options]
Note that you do not have the type the full command name; RosEasy is perfectly happy with the first word or even few letters of a command unless there is ambiguity.
setup_workspace <name> [-r]
Set up a new or remote workspace. This will walk you through the workspace
setup process. If setting up a remote workspace (i.e., on your own
computer for analysis), pass the -r option. This will set up a link between
the new workspace and the workspace on the cluster (make sure they are named
the same).
submit <workspace> <script path> [options]
Submits a job to be run on the cluster. This command takes a
python script as an argument. The only hard requirement is that the
python script defines which type of workspace Roseasy should create
for the task. The rest of the script may run a PyRosetta job, or simply
call RosettaScripts. It is recommended that you take an existing script
(found in standard_params) as a template, as there are a few lines at
the beginning of each script that handle figuring out which input file
goes with which task number and whether the script should be run in test
mode or not. You can also pass the --make-dirs option, which
causes the workspace to set up the sub-workspace without submitting any jobs.
This is useful for example when you want to handle step inputs manually.
Note that when the root directory is given as the argument,
RosEasy will automatically create a NEW sub-workspace, even if the previous
step is empty, so it is recommended that whenever possible you create the new
sub-workspace FIRST, either by passing --make-dirs or by running the
pick_designs command, and then pass that sub-workspace as the
argument.
generate_fragments <workspace> <step_number> [options]
Generates fragments for an FKIC run.
add_residues <pdb file or folder> <residue_string> <pdb_position>
Inserts residues into a PDB file and does a brief minimization
to close the chain break. You may specify any sequence to insert
using one-letter AA name formatting (i.e. 'SALTY'). The residues
will be inserted immediately after the <pdb_position> argument.
Defaults to chain A, pass the --chain option to specify a different
chain.
pick_designs_to_validate <step> [<picks_file>] [options]
Pick the designs that are at or near the Pareto front of the given metrics to
validate in the next step. You may also specify thresholds that designs much pass
in order to be chosen for the next step. This will create the workspace for the
next step and symlink the chosen input files, and is reccommended before any job.
For more information type 'roseasy pick_designs --help'
fetch_data <workspace>
On a "remote" workspace, fetch data from the main workspace.
push_data <workspace>
Push data from a "remote" workspace to the main workspace.
plot <directory> [options]
Generates a GUI for viewing designs.
First, set up a workspace on the cluster. It will ask you for a path to your Rosetta directory (the 'main' folder, rosetta/main in most installations), an input PDB file, and your Python executable where PyRosetta is installed. You will also be asked to provide a loop file, but you can ignore this if it is not applicable to your usage, or provide one later.
Cluster:
roseasy setup test
On your local workstation, set up the remote workspace.
Local:
roseasy setup test -r
Now, on the cluster, run your first script:
Cluster:
roseasy submit test test/standard_params/relax.py
When it's done running (check job status with the 'qstat' command), pull the data to your computer for analysis and to pick designs for the next step, then push your changes to the cluster.
Local:
roseasy fetch test
roseasy pick 2 test/standard_params/picks.yml
roseasy push test
Now you're ready to run the next step, in this example FKIC.
Cluster:
roseasy generate_fragments test/02
(wait for fragment generation to finish)
roseasy submit 02 standard_params/fkic.py
Pull your decoys to your local workstation and view them in the GUI.
Local:
roseasy fetch test
roseasy plot test/02/outputs/*
RosEasy is inspired by the code for Pull Into Place, developed by Kale Kundert and Cody Krivacic, and the desire to generalize its workspace paradigm to any Rosetta method.