Frequently asked questions¶

As you get started with emodpy-typhoid, you may have questions. The most common questions are answered below. The most common questions are answered below. For questions related to functionality in related packages, see the following documentation:

What are some of the key differences for people used to using dtk-tools?¶

Schema-Based. The creation of config and campaign files is entirely schema-based now. This means that you can only set parameters that the binary you are using recognizes. And parameter types and ranges are enforced at runtime.
Inter-File Dependencies Now Automatic. Before there were lots of parameters in the config that you had to set to correspond to settings in campaign or demographics files. That is no longer the case. We call these ‘implicits’. For example, if you add a BirthRate to the demographics, the corresponding parameters in the config.json (Enable_Births) will get set automatically for you. As another example, when you create a campaign and specify various ‘events’ to be broadcast/published and/or listened/subscribed to, you no longer have to figure out which ones are built-in and which are ad-hoc. It does that for you and populates the Custom_Events param on your behalf.
Hierarchical Dependencies Now Automatic. If a parameter depends on another parameter, previously you had to set all the Enables in the dependency tree. Now they get set automatically for you. For example, if Enable_Birth is set (see above), Enable_Vital_Dynamics will be set for you.
No JSON manipulation. dtk-tools worked primarily through manipulation of JSON that made up the configuration files. You no longer need to have any knowledge of the internal representation of data in the DTK input files. All configuration should be done via Python functions.
Released and Installed Modules. We want users mostly using versioned released modules that have been pip installed, not git cloned, dev-installed code, except during development. The process of getting new code reviewed, tested, and getting the module versioned and released is intended to be smooth and efficient when everyone does their defined role. “Trust The Process and Do Your Job”, as someone once said.
Blessed Binaries. In dtk-tools you would often BYOB – Bring Your Own Binary – but in emodpy, the idea is that the system pulls down the latest CI (Continuous Integration) build for your disease that passed all the tests. We very much want to noramlize the idea of doing research with versioned software that has come through our professional BVT processes.

Do I need to install a whole bunch of Python modules? Where do I get that list?¶

No. You should only have to install emodpy-typhoid, and everything else should come as a dependency automatically. If you are starting from a workflow repo, there should be a requirements.txt file that lists a particular version of emodpy-typhoid. Then ‘pip install -r requirements.txt’ is what you should expect to do.

Is there an easier way than typing –index-url with that long URL every time I use pip?¶

Yes. You should not be typing that every time. You should update the pip.ini (Windows) or pip.conf globally on your computer with that URL and never have to type it again.

How do I find the path or file to my pip.ini?¶

pip config -v list.

Choose the first or second one.

What if the pip.ini file (or pip folder) doesn’t exist?¶

Go ahead and create it (at one of the locations specified by ‘pip config -v list’). It’s your computer.

What’s the text I need to enter into my pip config file?¶

[global]
index-url = https://packages.idmod.org/api/pypi/pypi-production/simple

I was installing the Python modules and it told me I needed a C++ compiler?¶

This is because of a dependency in some versions of emod-api on lz4. lz4 usually comes as what’s called a source package and it has to be compiled as part of the install. This requires a compiler, which doesn’t come by default on Windows computers. At this point, if your pip.ini is set up properly, you should be able to get a pre-built lz4 package from our local Artifactory PIP server.

I was installing and got prompted for JFrog credentials.¶

You should not have to give credentials to use our JFrog/Artifactory/pip server. Credentials are required to get packages from the staging server. We don’t expect end users to be accessing packages from staging, just prod, which is auth-free.

What if I really actually do want to install something from staging?¶

You need to specify ‘–index-url = https://<username>@idmod.org:<shh…password>@packages.idmod.org/api/pypi/pypi-staging/simple’ and also provide your creds when prompted.

What version of Python should I be using?¶

At least Python 3.7.7. If you are installing a new version of Python, feel free to go all the way forward to a Python 3.9.x. 3.8 is probably the sweet spot of “known to work and still not considered old”.

What if I want a particular version of emodpy-typhoid?¶

pip install emodpy-typhoid==1.2.3

Should get you what you need.

What if I want to make changes to emodpy-typhoid and run with those, instead of a released version?¶

(This is a duplicate.)

There are a couple of ways of doing that. Option 1: Do a Dev Install:

pip install -e .

This will make your site packages map to your local code until you do a new pip install of the package.

Option 2: Creating a wheel from your local code and pip install it (each time you make a change).:

python setup.py bdist_wheel
pip3 install dist/<newly_create_file.whl>

Some people prefer option 1 because it’s “one and done”. Some people prefer option 2 because it keeps you thinking in terms of packaging, versioning, and installing even while you’re developing.

I pip installed emodpy-typhoid, but I want to make changes. How should I do that?¶

Install at a command prompt using the following:

python package_setup.py develop

This method is the most popular and proven, though there are some other options. Installing this way means that the emodpy-typhoid module in site-packages actually points to the same code as you have checked out in git. For more detail, see this Stack Overflow post.

However, we aim to get the desired changes quickly tested and included in the versioned module we release via pip install.

How do I set configuration parameters?¶

Define your own parameter-setting function such as set_param_fn and pass that function to the emodpy task creator as the param_custom_cb parameter. In that function, you can set the parameters directly. For example:

    print(f"Telling emod-api to use {manifest.schema_file} as schema.")
    camp.set_schema( manifest.schema_file )

    event = ob.new_intervention( camp, timestep=1, cases=1 )
    camp.add( event )
    return camp

def build_demog():
    """
    Build a demographics input file for the DTK using emod_api. 
    """
    import emodpy_typhoid.demographics.TyphoidDemographics as Demographics # OK to call into emod-api

    demog = Demographics.from_template_node( lat=0, lon=0, pop=10000, name=1, forced_id=1 )
    wb_births_df = pd.read_csv( manifest.world_bank_dataset )
    demog.SetEquilibriumVitalDynamicsFromWorldBank( wb_births_df=wb_births_df, country='Chile', year=2005 )
    return demog


def run_test():
    # Create a platform
    # Show how to dynamically set priority and node_group
    platform = Platform("SLURM", node_group="idm_48cores", priority="Highest") 

    task = EMODTask.from_default2(config_path="config.json", eradication_path=manifest.eradication_path, campaign_builder=build_camp, demog_builder=build_demog, schema_path=manifest.schema_file, param_custom_cb=set_param_fn, ep4_custom_cb=None)

    print("Adding asset dir...")

See examples/start_here/example.py. for additional information.

If you prefer something more modular, you can call a function in a standalone script/file that sets the configuration parameters.

Are there defaults?: Great question. If you don’t set any configuration parameters, they will have defaults based on the schema. The typhoid team has set team defaults in emodpy_typhoid.config.set_team_defaults(). These defaults can be seen in config.py.

How do I specify the log level for EMOD? I get a schema error when I try to set it now.¶

TBD

How do I specify the vector species for my scenario?¶

See the excerpt below or the complete example of setting the vector species and parameter values associated with each species in examples/start_here/example.py.

    print(f"Telling emod-api to use {manifest.schema_file} as schema.")
    camp.set_schema( manifest.schema_file )

    event = ob.new_intervention( camp, timestep=1, cases=1 )
    camp.add( event )
    return camp

def build_demog():
    """
    Build a demographics input file for the DTK using emod_api. 
    """
    import emodpy_typhoid.demographics.TyphoidDemographics as Demographics # OK to call into emod-api

    demog = Demographics.from_template_node( lat=0, lon=0, pop=10000, name=1, forced_id=1 )
    wb_births_df = pd.read_csv( manifest.world_bank_dataset )
    demog.SetEquilibriumVitalDynamicsFromWorldBank( wb_births_df=wb_births_df, country='Chile', year=2005 )
    return demog


def run_test():
    # Create a platform
    # Show how to dynamically set priority and node_group
    platform = Platform("SLURM", node_group="idm_48cores", priority="Highest") 

    task = EMODTask.from_default2(config_path="config.json", eradication_path=manifest.eradication_path, campaign_builder=build_camp, demog_builder=build_demog, schema_path=manifest.schema_file, param_custom_cb=set_param_fn, ep4_custom_cb=None)

    print("Adding asset dir...")

A helper function to make this task even easier may be coming shortly.

Where else should I search for functions?¶

Yes, emod-api. Any functionality that is not typhoid-specific (or disease-specific) will be found in emod-api. In particular you’ll probably find very useful functions for crafting campaigns in emod-api.interventions.common, such as the ScheduledCampaignEvent function and the TriggeredCampaignEvent function.

Do I need to be connected to the VPN?¶

The original way of procuring the model binary itself was via a call to get_model_files(). This required you to be VPN-ed in. This is no longer the preferred approach. Instead you will want to use the ‘bootstrap’ approach. This involves installing the emod_typhoid package, which should happen automatically, and using code like:

import emod_typhoid.bootstrap as dtk
dtk.setup(...)

This does not require VPN. The value you pass to setup is the path where the model files will be put.

Is there an example of creating a demographics file from scratch with the API?¶

Yes, see examples/create_demographics, there are also examples in emodpy-measles and emodpy-hiv. We are working to add some to emod_api.demographics. The basic idea is you use one of 3 node creators, and then use the Setter API to set up the node defaults for fertility, mortality, age structure, initial immunity, individual ‘risk’, and initial prevalance. The first node creator, from_template_node, is very basic and usually for quickstarts or toy models. It lets you create a single node demographics file with a given population. The second creator, from_csv, lets you create a multinode demographics using a csv file with population data as an input. The third creator, from_params, lets you create a multinode demographics without specific node data but instead with a few parameters that represent the overall population and the population heterogeneity.