MPFC & Ketamine Project

Overview

mPFC-ketamine Zotero Refs

DeepLabCut

Nath et al. (2019)
Mathis et al. (2020)
- DeepLabCut

DLCAnalyzer

Sturman, O, von Ziegler, L, Schläppi, C et al.(2020). Deep learning-based behavioral analysis reaches human accuracy and is capable of outperforming commercial solutions. Neuropsychopharmacology 45, 1942–52 https://doi.org/10.1038/s41386-020-0776-y
- Data from DLC (Python):wrangle/plot (R)

ETHZ-INS/DLCAnalyzer

Video links

Test out DLC (Basic)

https://www.mackenziemathislab.org/dlc-modelzoo
Single-frame analysis.
- This just works. It’s a small-scale version of what happens on a large-scale with all the 2d frames from your videos.

DLC Examples on Google Colab

Most of the DeepLabCut (DLC) tutorials are offered with Google Colab a cloud computing environment with all the software dependencies pre-installed. It really is the best first step if you want to start to understand how DLC works and how to customize it for your needs. - Click on the “Open in Colab” link on this page - Video on DLC on Google Colab - Colab instructions on how to use your own videos

A Worshop that has everything you need to use DLC

DeepLabCut Worshop Materials

HiPerGator Links

Some links for Getting started

YouTube video on using DeepLabCut GUI (~3 min)
Installing Deeplabcut https://deeplabcut.github.io/DeepLabCut/docs/installation.html
DLC User Guide https://deeplabcut.github.io/DeepLabCut/docs/standardDeepLabCut_UserGuide.html
Napari GUI https://deeplabcut.github.io/DeepLabCut/docs/napariGUI.html
Slightly outdated workshop https://github.com/DeepLabCut/DeepLabCut-Workshop-Materials

Video Behavior Analysis Workflow

%%{init: {'theme':'forest'}}%%

graph LR

    subgraph Analysis
       s6 -->  s7{Inferential Stats}
    end
 
    subgraph ETHZ-INS/DLCAnalyzer
        s3 --> s4[Data Wrangling </br> Graphing]
        s5 --> s6[Data Wrangling]
        s6 --> s4
        
        subgraph dlca[</tab> </br> Derivative Variables / Data Visualizations]
          s4 --> s5[unsupervised</br> classification</br> and clustering]
        end
    end

    subgraph DeepLabCut

        subgraph Training
            s1[add keypoints]-->s2[run training]
            s2 --> s2a[analyze</br> training]
            s2a -->s2

        end

        subgraph Decoding
            s2a --> s3a[add</br> videos] 
            
        end
        
         
        s3a --> s3[import data</br>transform]
       
    end
    
    subgraph Read-in-Data[Raw Data]
        s0[import video] --> s1
    end
    
 %% Notice that no text in shape are added here instead that is appended further down
    

    %% Comments after double percent signs

     classDef green fill:#9f6,stroke:#333,stroke-width:0.5px;
     classDef orange fill:#f96,stroke:#333,stroke-width:1px;
     classDef white fill:#fff,stroke:#333,stroke-width:1px;
     classDef sq stroke:#f66,stroke-width:1px;
     classDef blue fill:#6699cc,stroke:#333,stroke-width:1px;
     classDef red fill:#D32737,stroke:#FFF,stroke-width:1.5px;
     class sq,Analysis green
     class ETHZ-INS/DLCAnalyzer orange
     class Training,dlca,s0 white
     class DeepLabCut white
     class Decoding blue
     class Read-in-Data red

on Github/ Workshop

Unsupervised Clustering: (R) / (R Shiny)

0. Data Collection

1. Data Preprocessing

[ X ] Digitize behavior videos from analog
[ X ] Cut, Segment, Name: Subject & Session
[ X ] Extract behavior timestamp range
[ X ] Select behavior videos for DLC training

2. DeepLabCut Training

NEO/MNE e-phys

Mathis, Alexander, Steffen Schneider, Jessy Lauer, and Mackenzie Weygandt Mathis. 2020. “A Primer on Motion Capture with Deep Learning: Principles, Pitfalls, and Perspectives.” Neuron 108 (1): 44–65. https://doi.org/10.1016/j.neuron.2020.09.017.

Nath, Tanmay, Alexander Mathis, An Chi Chen, Amir Patel, Matthias Bethge, and Mackenzie Weygandt Mathis. 2019. “Using DeepLabCut for 3D Markerless Pose Estimation Across Species and Behaviors.” Nature Protocols 14 (7): 2152–76. https://doi.org/10.1038/s41596-019-0176-0.