Launch of our brand-new website for 𝐃𝐞𝐞𝐩𝐎𝐅! 🚀
👉 Explore: deepof.ai

DeepOF (Deep Open Field) processes DeepLabCut/SLEAP data using supervised and unsupervised annotation pipelines and using SHAP for interpretability.

Read about it in this month's post on OpenBehavior:
edspace.american.edu/openbehavior...

#OpenSource #Neuroscience #DeepLearning

New paper from the lab, where we show how the co-chaperone and psychiatric risk factor FKBP5 acts in CRH-expressing neurons to shape stress- sensitive social behavior. Deep phenotyping using #DLC and #DeepOF was key to unravel these effects. www.nature.com/articles/s41...

🖥️ Webinars
📆 Dec 12: @cost-teatime.bsky.social webinar with @mathiasvschmidt.bsky.social. Discover behavioral insights using the DeepOF tool—register now: helsinki.zoom.us/meeting/regi...
📆 Dec 2, 2024: @protocolsio.bsky.social shared tips on advancing reproducibility. Were you there? 🎥

🔍 What can you do with DeepOF?
𝘚𝘶𝘱𝘦𝘳𝘷𝘪𝘴𝘦𝘥 & 𝘜𝘯𝘴𝘶𝘱𝘦𝘳𝘷𝘪𝘴𝘦𝘥 𝘈𝘯𝘢𝘭𝘺𝘴𝘪𝘴: Detect pre-programmed behaviors (using built-in classifiers) or train machine learning models to discover entirely new, unbiased, unsupervised behavioral states.

Ever wondered if there is more to behavior than the location of the animal? Looking for a fast and easy way for in-depth behavioral analysis in any of your top-down experimental videos? Check out our package DeepOF
www.nature.com/articles/s41...

📺 Watch the Step-by-Step Setup Guide on YouTube: www.youtube.com/watch?v=bz6c...
Whether you are looking to scale up your behavioral screening, eliminate observer bias, or dive deep into the hidden structures of rodent behavior, DeepOF is built to streamline your workflow.

Check out the new website, grab the code, and let us know how DeepOF is supporting your research!

𝐃𝐞𝐞𝐩𝐎𝐅 bridges this gap.
Our open-source Python package offers an end-to-end pipeline for high-throughput, high-resolution deep phenotyping—fully accessible to the wider scientific community without requiring programming expertise.

What is 𝐃𝐞𝐞𝐩𝐎𝐅? Understanding animal behavior in high resolution is one of the greatest challenges in modern neuroscience. While deep learning tools have revolutionized pose estimation, translating the data into meaningful behavioral insights requires specialized setups and computational skills.

7) This work wouldn't have been possible without the contributions of the co-authors. Also a special thanks goes to the DeepLabCut development team www.mackenziemathislab.org/deeplabcut

And if you would like to use DeepOF check here:

deepof.readthedocs.io/en/latest/

4) 🚀 Using the open-source package DeepOF, we performed unsupervised analysis of fear behaviors. This revealed novel sex-specific patterns, in which females showed more active fear response (darting-like behavior) via explorative behaviors, that males did not show.

We developed a pipeline combining DeepLabCut, Keypoint-MoSeq & DeepOF to study age- and sex-dependent differences in light-tone second-order conditioning in mice.