DeepLabCut model for: Moreno-Gomez, M., et al. (2026). Off-target effects of DREADD ligands revealed by an anaesthesia emergence paradigm in mice.

This repository contains the trained DeepLabCut (v3.0.0rc6) pose estimation model used in the study: Off-target effects of DREADD ligands revealed by an anaesthesia emergence paradigm in mice (Moreno-Gomez M. et al., 2026).

The provided resources are part of an automated behavioural analysis pipeline designed to assess the off-target effects of chemogenetic ligands by measuring the dynamics of induction and emergence from isoflurane anaesthesia in mice.

DEEPLABCUT MODEL (POSE ESTIMATION AND TRACKING): The neural network is used to monitor the precise spatial coordinates of the animal throughout the experimental sessions.

├── Architecture: ResNet-50 with Group Normalization.
├── Training Dataset: 1,845 manually labeled frames from 75 different recordings.
├── Training Parameters: 200 epochs (batch size 8) using 95 percent of the labeled data.
└── Bodyparts Tracked (10): snout, head centre, left/right ears, neck, body centre, left/right body sides, tail base, and tail tip.

POST-TRACKING DATA ANALYSIS: The subsequent analysis of anaesthesia induction and emergence phases is performed using the raw coordinates generated by the tracking model.

├── Analysis Script: The complete pipeline for processing coordinates, identifying Walking onset through kinetic thresholds, and statistical evaluation is available on GitHub.
├── GitHub Repository: AnesthesiaDataAnalysis_MorenoGomez2026.
└── Validation: The automated pipeline results are highly correlated with expert manual scoring (r > 0.9).

INCLUDED FILES

├── config.yaml                   (DeepLabCut project configuration and metadata)
├── pytorch_config.yaml     (Configuration for the PyTorch-based network)
├── snapshot-200.pt           (Trained neural network weights)
└── learning_stats.csv         (Log of the training performance)

HOW TO CITE

Please cite the original paper: Moreno-Gomez, M., et al. (2026). Off-target effects of DREADD ligands revealed by an anaesthesia emergence paradigm in mice.