marm-viz 0.2.0

Visualization toolkit for marmoset homecage behavioral data (marm_behavior outputs).

marm_viz

Visualization toolkit for marmoset homecage behavioral data. Designed to work with outputs from marm_behavior.

Installation

pip install marm-viz

For the interactive annotation tools (cage boundaries, stimulus centre), install with the annotate extra:

pip install 'marm-viz[annotate]'

Quick Start

Point marm_viz at a folder containing marm_behavior outputs (hlabel, hcoord, edges, depths files):

# Generate all 10 plot types (default: JPG output)
python -m marm_viz /path/to/recording_folder

# Single plot type
python -m marm_viz /path/to/recording_folder --plot ethogram

# Vector output for Illustrator
python -m marm_viz /path/to/recording_folder --format svg

Plot Types

Flag	Output	Description
qc	QC panels	8-panel overview: head XY colored by depth/distance/angle/behavior + histograms
heatmaps	Category heatmaps	Per-category spatial occupancy + near-stimulus + overall occupancy + P(move)
timeseries	Stim proximity	Fraction near stimulus across time chunks + per-chunk heatmaps + pooled XY
density	2D density	Smoothed occupancy density with contours
behavior	State usage + t-SNE	80-state bar chart grouped by category + t-SNE centroid bubble plot
ethogram	Ethogram	Per-category binary raster for each animal across time
transitions	Transition matrix	5x5 category transition probabilities (diagonal NaN'd)
polar	Head direction	Polar histogram of head angle relative to stimulus (near vs far)
track	Head tracks	Per-animal + merged XY trajectory line plots
correlation	Cagemate correlation	Pairwise Spearman correlation of 80-state usage vectors across time chunks

CLI Reference

Mode Selection

If neither --plot nor --annotate is given, all 10 plot types are generated.

Flag	Description
--plot TYPE	Generate a single plot type (see table above)
--annotate MODE	Interactive annotation: metadata, boundaries, center, or all

Output

Flag	Default	Description
--output-dir DIR, -o	<folder>/marm_viz_output/	Output directory (created if missing)
--format FMT	jpg	Output format: jpg, svg, png, or pdf

Recording Selection

Flag	Default	Description
--video STEM	all	Process only this video stem (e.g. test_4)
--color {r,w,b,y}	all	Process only this animal colour
--animal-to-use COLOR	all	Same as --color but also accepts full names (Red, White, Blue, Yellow)
--type-to-use STR	all	Only process videos whose filename contains STR (case-insensitive)
--time-to-use MINUTES	all	Limit each recording to this many minutes of data
--state-remap FILE	bundled	Path to a custom state-reorder file (80 ints, one per line)

Plotting Parameters

Flag	Default	Description
--fps N	60	Frames per second
--bin-size N	20	Bin size in pixels for heatmap plots
--radius-px N	120	Near-stimulus radius in pixels
--chunk-minutes N	5	Chunk length in minutes for time series
--n-frames N	50000	Max frames for QC scatter plots (0 = use all)
--scatter-sample-rate F	0.01	Fraction of points for scatter subplots (1%)
--mid-color {purple,green}	purple	Colormap middle colour for density plots
--log-alpha N	10	Log tone-map aggressiveness for density (0 = linear)
--use-body-xy	off	Use body centroid instead of head position
--augment	off	Left/right flip augmentation (doubles data, removes spatial bias)

Annotation Parameters

Flag	Default	Description
--force	off	Re-annotate files that already exist
--id-parent-dir DIR	parent of folder	Where animal_ID.txt should live

General

Flag	Description
-v, --verbose	Print detailed diagnostic information
-h, --help	Show help and exit

Python API

from marm_viz import build_recording_set, plot_ethogram

data = build_recording_set("/path/to/folder")
fig = plot_ethogram(data, fps=60)
fig.savefig("ethogram.svg")

For detailed documentation on each plot function and the Recording data structure, see the User Guide.

License

MIT