Functional brain connectivity fingerprints identify polytherapy candidates. a Schematic representation of the light stimulus parameters and associated activity states used for functional connectivity fingerprinting. GCaMP activity is recorded for 10 min in the absence of light stimuli (pre-stimulus state). Seizure-inducing white light stimuli are then applied every two minutes. GCaMP recordings during this period are subdivided into early post-stimulus (0–60 s after the stimulus) and late post-stimulus (60–120 s after the stimulus) states during subsequent analysis. b Representative local field potential recordings (LFP traces; upper panels) and single frames from brain-wide GCaMP recordings (GCaMP activity; lower panels) in scn1lab mutants (top) and wild-type controls (bottom) illustrating response to light stimuli at 5 dpf. LFP trances and GCaMP activity are recorded independently. Gray bands on LFP recordings indicate the timing of the two 500 ms light pulses. Scale bars show 0.5 s (x-axis) and 1 mV (y-axis). GCaMP images show activity from a single z-plane 5 s before the first pulse (pre-stimulus) and again 5 s after the second pulse (post-stimulus). Supervoxels with significant changes in fluorescence activity (∆F/F) are shown in color. c For each compound, a 165-metric functional connectivity fingerprint is generated based on correlation coefficients between active supervoxels in all 55 pairs of brain areas during each of the three activity states. Each square represents the deviation of the correlation for the compound in question (X) from the mean correlation for all compounds (X_mean) in the brain area pair being analyzed, normalized by standard deviation [(X−X_mean)/X_STD]. The analysis is carried out in both mutants (Mut, magenta text) and sibling controls (WT, cyan text) prior to compound addition (0 h) and again at 4 h post-exposure to the compounds in 1 % DMSO (4h). Fingerprints are analyzed by hierarchical clustering to identify compounds that modulate functional connectivity networks in similar ways. We identify the three most distinct clusters (C1, C2, and C3) that fall closest to the wild-type cluster. We then select the compound in each activity cluster that most effectively restores functional connectivity in scn1lab mutants based on Euclidean distance from the wild-type cluster (indicated by checkmarks). Source data are provided as a Source Data file