Home
Welcome to the MoSeq2 Wiki! MoSeq is a tool to extract mouse pose from depth video and model how pose evolves over time. The documentation should reflect the latest version of MoSeq2. If it does not, please submit an issue describing where the documentation needs updating. Please join our Slack channel! Once you join you'll be able to talk to the MoSeq team directly for help or information and talk to other community members about how you use MoSeq.
Start here to learn more about how to use MoSeq for your experiments. Please read through these guides before setting up MoSeq to gain familiarity with the pipeline. Use the sidebar to navigate through the different topics related to MoSeq. You can find a nice high-level overview of MoSeq in our README. This wiki contains guides for:
- Hardware setup for data acquisition
- Documentation for data acquisition
- Steps for MoSeq2 installation
- Guides for running the MoSeq2 pipeline
- Upgrading MoSeq
Many of MoSeq's interactive features require using a Jupyter notebook or JupyterLab. Please take some time to get familiar with these tools.
Read below for an overview of MoSeq2's documentation. You can find links for navigating the wiki in the sidebar too.
Note: the current version of MoSeq works with depth video only. We hope to support other data types, such as keypoint tracking, in the future.
Head to the experiment setup page for links to
- the bill of materials for the acquisition cage
- installation instructions
- recording bucket preparation
Click the following link for instructions on how to use the acquisition software with the Kinect v2.
We are currently working on testing the Azure Kinect acquistion pipeline and you can find the beta version and the documentation here:
There are 2 options for using MoSeq:
- Docker (recommended installation route)
- Anaconda (recommended for experienced python users or HPC/SLURM users)
Pick out the option that is most suitable for your needs. However, we strongly urge first-time users to install MoSeq2 via Docker. You can also find this information on the MoSeq2 installation page.
We strongly suggest this option for people new to using the terminal or new to programming. Docker runs virtual containers with MoSeq2 pre-installed. This means that once Docker is installed and you download the MoSeq2 Docker image, you'll be able to start using MoSeq2 immediately. Use the following links for platform-specific installation steps:
We recommend the Anaconda installation route for users who have more experience with the command line, python, and virtual environments or users using a high-performance computing cluster.
This installation route uses the Anaconda environment and package manager to install a MoSeq2 environment. Users have more flexibility with this method.
Links for platform-specific installation steps are found here:
After successfully installing MoSeq2, head on over to the pipeline documentation to get familiar with the two main components of the MoSeq2 pipeline:
- extraction and modeling: via Jupyter notebooks or the command-line interface
- analysis: via Jupyter notebooks or the command-line interface
There are two routes for upgrading.
- Navigate into the
moseq2-apprepository and pull the latest release withgit pull origin release. Then runpip install .to re-installmoseq2-app. - Create a new Conda environment and run through the Conda installation steps from start to finish.
NOTE: Run route 1 with caution, as there is a slight possibility that it will mess up your current moseq2 Conda environment.
If that happens, you'll have to delete your current environment any try the Conda installation steps again.
Route 2 is the safer option.
Rename the old Jupyter notebooks so you don't accidentally overwrite them, then copy the notebooks in the moseq2-app repository
To upgrade via Docker, look for the most recent Docker image under the dattalab remote repository and download it.
Set up a new container form this image and link it to your project folder.
See the Docker installation steps for more details.
If you want to update the MoSeq2 Jupyter notebooks, you must rename the notebooks currently in your project folder or move them in a folder. Running the Docker container will automatically copy the new notebooks over to your project folder only if it doesn't find files with the same name.
Other tutorials:
Resources:
Latest versions
moseq2-extract==1.1.2moseq2-pca==1.1.3moseq2-model==1.1.2moseq2-viz==1.2.0moseq2-app==1.2.0
MoSeq2 logo by Sigrid Knemeyer
Home | Changelog | Acquisition | Installation | Analysis Pipeline | Troubleshooting and Tips |
Tutorials |
Join our slack
