Tutorial 2: Using CAP#

colab

The CAP class is designed to perform CAPs analyses (on all subjects or group of subjects). It offers the flexibility to analyze data from all subjects or focus on specific groups, compute CAP-specific metrics, and generate visualizations to aid in the interpretation of results.

Performing CAPs on All Subjects#

All information pertaining to CAPs (k-means models, activation vectors/cluster centroids, etc) are stored as attributes in the CAP class and this information is used by all methods in the class. These attributes are accessible via properties. Some properties can also be used as setters.

import numpy as np
from neurocaps.analysis import CAP

# Extracting timseries
parcel_approach = {"Schaefer": {"n_rois": 100, "yeo_networks": 7, "resolution_mm": 2}}

# Simulate data for example; Subject IDs will be sorted lexicographically
sub_ids = [f"0{x}" if x < 10 else x for x in range(1, 11)]
subject_timeseries = {str(x): {f"run-{y}": np.random.rand(100, 100) for y in range(1, 4)} for x in sub_ids}

# Initialize CAP class
cap_analysis = CAP(parcel_approach=parcel_approach)

# Get CAPs
cap_analysis.get_caps(
    subject_timeseries=subject_timeseries,
    n_clusters=range(2, 11),
    cluster_selection_method="elbow",
    show_figs=True,
    step=2,
    progress_bar=True,  # Available in versions >= 0.21.5
)

2025-04-15 18:14:40,114 neurocaps.analysis.cap [INFO] No groups specified. Using default group 'All Subjects' containing all subject IDs from `subject_timeseries`. The `self.groups` dictionary will remain fixed unless the `CAP` class is re-initialized.
Collecting Subject Timeseries Data [GROUP: All Subjects]: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10/10 [00:01<00:00, 668.15it/s]
Concatenating Timeseries Data Per Group: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10/10 [00:01<00:00, 178.01it/s]
Clustering [GROUP: All Subjects]: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 9/9 [00:00<00:00, 20.38it/s]
2025-04-15 18:15:07,367 neurocaps.analysis.cap [INFO] [GROUP: All Subjects | METHOD: elbow] Optimal cluster size is 5.
../_images/All_Subjects_elbow.png

print can be used to return a string representation of the CAP class.

print(cap_analysis)

Current Object State:
================================================
Parcellation Approach                                      : Schaefer
Groups                                                     : All Subjects
Number of Clusters                                         : [2, 3, 4, 5, 6, 7, 8, 9, 10]
Cluster Selection Method                                   : elbow
Optimal Number of Clusters (if Range of Clusters Provided) : {'All Subjects': np.int64(5)}
CPU Cores Used for Clustering (Multiprocessing)            : None
User-Specified Runs IDs Used for Clustering                : None
Concatenated Timeseries Bytes                              : 2400184 bytes
Standardized Concatenated Timeseries                       : True
Co-Activation Patterns (CAPs)                              : {'All Subjects': 5}
Variance Explained by Clustering                           : {'All Subjects': np.float64(0.02448526803307005)}

Performing CAPs on Groups#

cap_analysis = CAP(groups={"A": ["01", "02", "03", "05"], "B": ["04", "06", "07", "08", "09", "10"]})

cap_analysis.get_caps(
    subject_timeseries=subject_timeseries,
    n_clusters=range(2, 21),
    cluster_selection_method="silhouette",
    show_figs=True,
    step=2,
    progress_bar=True,
)

Collecting Subject Timeseries Data [GROUP: A]: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4/4 [00:01<00:00, 582.04it/s]
Collecting Subject Timeseries Data [GROUP: B]: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:01<00:00, 706.37it/s]
Concatenating Timeseries Data Per Group: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:01<00:00, 308.08it/s]

Clustering [GROUP: A]: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 19/19 [00:01<00:00, 18.71it/s]
2025-04-15 18:15:53,981 neurocaps.analysis.cap [INFO] [GROUP: A | METHOD: silhouette] Optimal cluster size is 2.
../_images/A_silhouette.png 
Clustering [GROUP: B]: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 19/19 [00:01<00:00, 12.48it/s]
2025-04-15 18:15:55,236 neurocaps.analysis.cap [INFO] [GROUP: B | METHOD: silhouette] Optimal cluster size is 2.
../_images/B_silhouette.png

Calculate Metrics#

Note that if standardize was set to True in CAP.get_caps(), then the column (ROI) means and standard deviations computed from the concatenated data used to obtain the CAPs are also used to standardize each subject in the timeseries data inputted into CAP.calculate_metrics(). This ensures proper CAP assignments for each subjects frames.

df_dict = cap_analysis.calculate_metrics(
    subject_timeseries=subject_timeseries,
    return_df=True,
    metrics=["temporal_fraction", "counts", "transition_probability"],
    continuous_runs=True,
    progress_bar=True,
)

print(df_dict["temporal_fraction"])

Computing Metrics for Subjects: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10/10 [00:00<00:00, 159.78it/s]

Subject_ID

Group

Run

CAP-1

CAP-2

01

A

run-continuous

0.5066666666666667

0.49333333333333335

02

A

run-continuous

0.5333333333333333

0.4666666666666667

03

A

run-continuous

0.6

0.4

05

A

run-continuous

0.54

0.46

04

B

run-continuous

0.41333333333333333

0.5866666666666667

06

B

run-continuous

0.47333333333333333

0.5266666666666666

07

B

run-continuous

0.44

0.56

08

B

run-continuous

0.5

0.5

09

B

run-continuous

0.4866666666666667

0.5133333333333333

10

B

run-continuous

0.46

0.54

Plotting CAPs#

import seaborn as sns

cap_analysis = CAP(parcel_approach={"Schaefer": {"n_rois": 100, "yeo_networks": 7, "resolution_mm": 1}})

cap_analysis.get_caps(subject_timeseries=subject_timeseries, n_clusters=6)

sns.diverging_palette(145, 300, s=60, as_cmap=True)

palette = sns.diverging_palette(260, 10, s=80, l=55, n=256, as_cmap=True)

kwargs = {
    "subplots": True,
    "fontsize": 14,
    "ncol": 3,
    "sharey": True,
    "tight_layout": False,
    "xlabel_rotation": 0,
    "hspace": 0.3,
    "cmap": palette,
}

cap_analysis.caps2plot(visual_scope="regions", plot_options="outer_product", show_figs=True, **kwargs)

2025-04-15 18:16:21,487 neurocaps.analysis.cap [INFO] No groups specified. Using default group 'All Subjects' containing all subject IDs from `subject_timeseries`. The `self.groups` dictionary will remain fixed unless the `CAP` class is re-initialized.
../_images/All_Subjects_CAPs_outer_product_heatmap-regions.png 
cap_analysis.caps2plot(
    visual_scope="nodes", plot_options="heatmap", xticklabels_size=7, yticklabels_size=7, show_figs=True
)
../_images/All_Subjects_CAPs_heatmap-nodes.png

Generate Pearson Correlation Matrix#

cap_analysis.caps2corr(annot=True, cmap="viridis", show_figs=True)
../_images/All_Subjects_CAPs_correlation_matrix.png 
corr_dict = cap_analysis.caps2corr(return_df=True)
print(corr_dict["All Subjects"])

CAP-1

CAP-2

CAP-3

CAP-4

CAP-5

CAP-6

CAP-1

1 (0)***

-0.24 (0.016)*

-0.26 (0.01)*

-0.1 (0.3)

-0.17 (0.087)

-0.17 (0.09)

CAP-2

-0.24 (0.016)*

1 (0)***

-0.11 (0.28)

-0.15 (0.14)

-0.28 (0.0051)**

-0.28 (0.0055)**

CAP-3

-0.26 (0.01)*

-0.11 (0.28)

1 (0)***

-0.3 (0.0021)**

-0.18 (0.075)

-0.19 (0.058)

CAP-4

-0.1 (0.3)

-0.15 (0.14)

-0.3 (0.0021)**

1 (0)***

-0.18 (0.076)

-0.22 (0.028)*

CAP-5

-0.17 (0.087)

-0.28 (0.0051)**

-0.18 (0.075)

-0.18 (0.076)

1 (0)***

-0.17 (0.089)

CAP-6

-0.17 (0.09)

-0.28 (0.0055)**

-0.19 (0.058)

-0.22 (0.028)*

-0.17 (0.089)

1 (0)***

Creating Surface Plots#

from matplotlib.colors import LinearSegmentedColormap

# Create the colormap
colors = [
    "#1bfffe",
    "#00ccff",
    "#0099ff",
    "#0066ff",
    "#0033ff",
    "#c4c4c4",
    "#ff6666",
    "#ff3333",
    "#FF0000",
    "#ffcc00",
    "#FFFF00",
]

custom_cmap = LinearSegmentedColormap.from_list("custom_cold_hot", colors, N=256)

# Apply custom cmap to surface plots
cap_analysis.caps2surf(progress_bar=True, cmap=custom_cmap, size=(500, 100), layout="row")

Generating Surface Plots [GROUP: All Subjects]: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:07<00:00,  3.91s/it]
../_images/All_Subjects_CAP-1_surface_plot.png ../_images/All_Subjects_CAP-2_surface_plot.png

Plotting CAPs to Radar#

radialaxis = {
    "showline": True,
    "linewidth": 2,
    "linecolor": "rgba(0, 0, 0, 0.25)",
    "gridcolor": "rgba(0, 0, 0, 0.25)",
    "ticks": "outside",
    "tickfont": {"size": 14, "color": "black"},
    "range": [0, 0.6],
    "tickvals": [0.1, "", "", 0.4, "", "", 0.6],
}

legend = {
    "yanchor": "top",
    "y": 0.99,
    "x": 0.99,
    "title_font_family": "Times New Roman",
    "font": {"size": 12, "color": "black"},
}

colors = {"High Amplitude": "red", "Low Amplitude": "blue"}

kwargs = {
    "radialaxis": radialaxis,
    "fill": "toself",
    "legend": legend,
    "color_discrete_map": colors,
    "height": 400,
    "width": 600,
}

cap_analysis.caps2radar(**kwargs)
../_images/All_Subjects_CAP-1_radar.png ../_images/All_Subjects_CAP-2_radar.png

Download Jupyter Notebook