Getting started


The CLAMS project has many moving parts to make various computational analysis tools talk to each other to create customized workflow pipelines. However the most important part of the project must be the apps published for the CLAMS platform. The CLAMS Python SDK will help app developers handling MMIF data format with high-level classes and methods in Python, and publishing their code as a CLAMS app that can be easily deployed to the site via CLAMS workflow engines, such as the CLAMS appliance.

A CLAMS app can be any software that performs automated contents analysis on text, audio, and/or image/video data stream. An app must use MMIF as I/O format. When deployed into a CLAMS appliance, an app needs be running as a webapp wrapped in a docker container. In this documentation, we will explain what Python API’s and HTTP API’s an app must implement.


  • Python: clams-python requires Python 3.6 or newer. We have no plan to support Python 2.7.

  • Docker: when deployed to the CLAMS appliance, an app must be in a docker container


clams-python distribution package is available at PyPI. You can use pip to install the latest version.

pip install clams-python

Note that installing clams-python will also install mmif-python PyPI package, which is a companion python library related to the MMIF data format.

I/O Specification

A CLAMS app must be able to take a MMIF json as input as well as to return a MMIF json as output. MMIF is a JSON(-LD)-based open source data format. For more details and discussions, please visit the MMIF website and the issue tracker.

mmif package

mmif-python PyPI package comes together with the installation of clams-python, and with it, you can use mmif python package.

import mmif
from mmif.serialize import Mmif
new_mmif = Mmif()
# this will fail because an empty MMIF will fail to validate against MMIF JSON schema

Because API’s of the mmf package is well documented in the mmif-python website, we won’t go into more details here. Please refer to the website.

Note on versions

clams-python is under active development, so is mmif-python, which is a separate PyPI distribution package providing Python classes and methods to handle MMIF JSON string. Because of this rapid version cycles, it is possible that a MMIF file of a specific version does not work with a CLAMS app that is based on a incompatible version of mmif-python. In every MMIF files, there must be the MMIF version encoded at the top of the file. Please keep in mind the versions of Python libraries you’re using and their target specification version. To see the MMIF specification version targeted by the installed mmif-python package, look at mmif.__specver__ variable (installing clams-python will also install mmif-python).

import mmif

A CLAMS app must report which MMIF specification version it targets in its metadata (see CLAMS App Metadata). And when an app targets a specific version, it means;

  1. the app can only process input MMIF files that are compatible with the target version.

  2. the app will output MMIF files exactly versioned as the target version.

Finally, for two different specifications to be compatible to each other, their major and minor version numbers should be the same. For example, 0.4.2 is compatible either with 0.4.10 or 0.4.0, but is not compatible with 0.3.0 or 0.7.2.

For more information on the relation between mmif-python versions and MMIF specification versions, or MMIF version compatibility, please take time to read our decision on the subject here. You can also find a table with all public clams-python packages and their target MMIF versions in Target MMIF Versions. This is a seemingly complicated issue, but also is very crucial to build CLAMS as a platform.


A CLAMS Python app is a python class that implements and exposes two core methods; annotate(), appmetadata().

  • appmetadata(): Returns JSON-formatted str that contains metadata about the app.

  • annotate(): Takes a MMIF as the only input and processes the MMIF input, then returns serialized MMIF str.

A good place to start writing a CLAMS app is to start with inheriting And if you’re doing so, you might want to implement two private methods instead of two public methods above. That’s because the implementation of the public methods in the super class internally call these private methods respectively.

We provide a tutorial for writing with a real world example at <Tutorial: writing a CLAMS app>. We highly recommend you to go through it.

Note on App metadata

App metadata is a map where important information about the app itself is stored as key-value pairs. See <CLAMS App Metadata> for the specification. In the future the app metadata will be used for automatic generation of CLAMS App index in the CLAMS App Directory, as well as automatic integration to Galaxy in the appliance deployment.

HTTP webapp

To be integrated into the CLAMS appliance, a CLAMS app needs to serve as a webapp. Once your application class is ready, you can use clams.restify.Restifier to wrap your app as a Flask-based web application.

from import ClamsApp
from clams.restify import Restifier

class AnApp(ClamsApp):
    # Implements an app that does this and that.
    # Must implement `_appmetadata`, `_annotate` methods

if __name__ == "__main__":
    app = AnApp()
    webapp = Restifier(app)

When running the above code, Python will start a web server and host your CLAMS app. By default the serve will listen to, but you can adjust hostname and port number. In this webapp, appmetadata and annotate will be respectively mapped to GET, and POST to the root route. Hence, for example, you can POST a MMIF file to the web app and get a response with the annotated MMIF string in the body.

In the above example, will start the webapp in debug mode on a Werkzeug server, which is not always suitable for a production server. For a more robust server that can handle multiple requests asynchronously, you might want to use a production-ready HTTP server. In such a case you can use serve_production(), which will spin up a multi-worker Gunicorn server. If you don’t like it (for example, gunicorn does not support Windows OS), you can write your own HTTP wrapper. In the end of the day, all you need is a webapp that maps appmetadata and annotate on GET and POST requests.


In addition to the HTTP service, a CLAMS app is expected to be containerized. Concretely, the appliance maker expects a CLAMS app to have a Dockerfile at the project root. Independently from being compatible with the CLAMS appliance, containerization of your app is recommended especially when your app processes video streams and dependent on complicated system-level video processing libraries (e.g. OpenCV, FFmpeg).

Refer to the official documentation to learn how to write a Dockerfile. To integrate to the CLAMS appliance, a dockerized CLAMS app must automatically start itself as a webapp when instantiated as a container, and listen to 5000 port in the container.

We have a public docker hub, and publishing Debian-based base images to help developers write Dockerfile and save build time to install common libraries. At the moment we have a basic image with Python 3.6 and clams-python installed. We will publish more images built with commonly used video and audio processing libraries.

CLAMS appliance integration

Finally, here are requirements for an app to be appliance compatible.

  1. App code is hosted on a public git repository.

  2. App is dockerized

  3. The app docker image will automatically start the app as a webapp, and listen to port 5000.

  4. Dockerfile for the dockerization is placed in the root of the git repository

To learn how to deploy your app on an appliance instance, please refer to the appliance documentation.