ModelManager

manage and run models in queue

usage

./rq_starter {worker_name}

INSTALL

step 1:

pip install -U -r requirenments.txt --process-dependency-links --allow-all-external

step 2:

read CIMtools installation guide.

step 3:

add MWUI

edit ~/.CIMtools.ini or add /etc/.CIMtools.ini or add file into package CIMtools/.CIMtools.ini [useful for unpacked MODtools Lib]

priority: CIMtools/.CIMtools.ini >> ~/.CIMtools.ini >> /etc/.CIMtools.ini

Intro

MWUI server manage access to models which can be stored on several machines.

Modeler server

Machine with redis server available from network. redis used for storing tasks to queue and after modeling done for results temporary storing.

Configuration

modeler server have to following programs:

• python 3.4 or 3.5 or 3.6
• python modules from requirements.txt
• MWUI, CGRtools and MODtools libs
• Fragmentor binaries
• Colorize and over utils
• ChemAxon Jchem
• Chemaxon REST (optional)

Fragmentor binaries have to stored in directory with name format: fragmentor-{version}

MWUI and MODtools must be configured. Need correct paths to binaries.

Model attachment

Available two dirs for common model storing.

• alienmodel - for storing various models in separate subdirs.
• modelbuilder - for storing models autogenerated with modelbuilder script

Nonstandard models have to be written on Python and stored here as *.py script

Nonstandard models

script have to contain ModelLoader class with methods:

load_model(name) - return model object for given model name

get_models() - return list of models available in script in format:

[{"example": "smiles or cml or mrv",
  "description": "description if Markdown format",
  "type"="ModelType object", "name": "model name showed in MWUI"}]

model object contain methods:

set_work_path(path) -