BehaviorDEPOT is a simple, flexible tool for automated behavioral detection based on markerless pose tracking
- Department of Physiology, University of California, Los Angeles, United States
- UCLA Neuroscience Interdepartmental Program, University of California, Los Angeles, United States
- UCLA Molecular, Cellular, and Integrative Physiology Program, University of California, Los Angeles, United States
- Department of Neurology, University of California, Los Angeles, United States
- Department of Psychiatry, University of California, Los Angeles, United States
- Department of Psychology, University of California, Los Angeles, United States
Figures
Figure 1 with 1 supplement
BehaviorDEPOT is a general-purpose behavioral analysis software comprising six modules.
The Experiment Module is a MATLAB application with a graphical interface that allows users to design and run fear conditioning experiments. The software uses Arduinos to interface with commercially …
Figure 1—figure supplement 1
Example arrangement of Arduino interface between computer, fear conditioning, and optogenetics hardware.
The Experiment Module controls two Arduinos that control delivery of the scrambled shocker, and a light (for use as a conditioned cue), and the laser for optogenetics, respectively. MATLAB software …
Figure 2 with 2 supplements
The Analysis Module.
(A) The Analysis Module workflow. Videos and accompanying pose tracking data are the inputs. Pose tracking and behavioral data is vectorized and saved in MATLAB structures to facilitate subsequent …
Figure 2—figure supplement 1
Performance of the freezing heuristic based on DLC mean tracking error.
Heuristic performance statistics plotted against root mean squared error (RMSE) of the DLC model. N=6 videos were tested to generate average heuristic performance for each model. Error bars, S.E.M. …
Figure 2—figure supplement 2
Performance of the ‘Jitter’ Freezing Heuristic on Webcam videos.
(A) Human vs. velocity vs. jitter freezing annotations (F(1.4,4.2)=0.32, P=0.67, RM one-way ANOVA). (B) Evaluation of freezing heuristic performance on videos recorded at 30fps with a standard …
Figure 3 with 1 supplement
Use Case 1: Optogenetics.
(A) AAV1-CamKII-GtACR2-FusionRed was injected bilaterally into medial prefrontal cortex (mPFC). (B) Behavioral protocol. Mice underwent contextual fear conditioning on day 1. On day 2, mice were …
Figure 3—figure supplement 1
Histology for optogenetics viral injections and fiber implants.
(A) Optic fiber cannula placements for experiment described in Figure 3. (B) StGtACR2 -FusionRed expression and bilateral fiber placement for representative shocked and non-shocked mice. Scale bar, …
Figure 4
Use Case 2: Mice wearing Miniscopes.
(A) Design for MiniCAM, an open-source camera designed to interface with Miniscopes and pose tracking. (B) Still frame from MiniCAM recording of mouse wearing a V4 Minscope. DLC tracked points are …
Figure 5
Use cases 3–5: EPM, OFT, NOE.
(A–C) Screens shot from Analysis Module showing user-defined ROIs in the EPM, OFT and NOE. Scale bars, 10 cm. (D) Statistical comparison of human vs. BehaviorDEPOT ratings for time in each arm (FRate…
Figure 6
Use Case 6: Automated analysis of an effort-based decision-making T-maze task.
(A) Screen shots showing DLC tracking in a one-barrier (top) and two-barrier (bottom) T-maze and ROIs used for analysis in BehaviorDEPOT. (B) Sample mouse trajectories in a one-barrier (top) and …
Figure 7
Sample outputs of the Inter-Rater Module.
(A) The Inter-Rater Module imports reference annotations, converts them to a BehaviorDEPOT-friendly format, aligns annotations, and reports statistics about agreement between raters. (B1) Alignment …
Figure 8
The Data Exploration Module.
(A) The Data Exploration Module takes in metrics from the Analysis Module along with reference annotations. It sorts the data, separating frames containing the behavior of interest from those …
Figure 9
Analysis of External Data using Optimization and Validation Modules.
(A) Optimization Module workflow. This module sweeps through a range of thresholds for metrics calculated based on tracked points and then compares the resulting behavioral classification to human …
Figure 10
Comparisons with JAABA.
(A) MoTr tracking in a standard fear conditioning chamber (left) and an open field (right). (B) Ctrax tracking in a standard fear conditioning chamber (left) and an open field (right). (C) DLC …
Additional files
-
Supplementary file 1
List and descriptions of tracked keypoints and keypoints calculated by the Analysis Module.
- https://cdn.elifesciences.org/articles/74314/elife-74314-supp1-v1.xlsx
-
Supplementary file 2
List and descriptions of metrics calculated by the Analysis Module.
- https://cdn.elifesciences.org/articles/74314/elife-74314-supp2-v1.xlsx
-
Supplementary file 3
List of required inputs and automatically generated outputs for every module.
- https://cdn.elifesciences.org/articles/74314/elife-74314-supp3-v1.xlsx
-
Supplementary file 4
Descriptions of error rates in DLC keypoint tracking networks and descriptions of held-out video sets used to test each heuristic.
- https://cdn.elifesciences.org/articles/74314/elife-74314-supp4-v1.xlsx
-
Supplementary file 5
Descriptions of DLC models used to assess the relationship between DLC mean error and freezing heuristic performance.
- https://cdn.elifesciences.org/articles/74314/elife-74314-supp5-v1.xlsx
-
MDAR checklist
- https://cdn.elifesciences.org/articles/74314/elife-74314-mdarchecklist1-v1.docx
