Fig. 3: Matrices for the differences in functional network connectivity between controls (N = 626) and cognitive biotypes.

Discovery set: biotype 1, N = 426; biotype 2, N = 497. Replication set: biotype 1, N = 110, biotype 2 N = 146. First-degree relatives: biotype 1, N = 153; biotype 2, N = 178, control biotype, N = 95. For each functional network connectivity (FNC) feature, we fit a linear model adjusting for sex, age, race, ethnicity, site, and head motion and conducted a two-tailed t-test to compare controls with each biotype of patients and relatives. Bottom-left triangle: t-statistic. Top-right triangle: p-values, those that reached statistical significance (p < 0.05) are shown with colors (–log10(p-value)×sign(t-statistic) scale); otherwise, they are shown in gray. The discovery set is corrected for multiple comparisons (False Discovery Rate). Yellow-Red colors: higher FNC in patients/relatives compared to controls. Blue colors: lower FNC in patients/relatives compared to controls. A Biotype 1 from discovery set. B Biotype 2 from discovery set. C Biotype 1 from replication set. D Biotype 2 from replication set. E First-degree relatives biotype 1. F First-degree relatives biotype 2. G First-degree relatives control biotype. Cognitive FNC features in psychosis (CFPs) included in k-means clustering are shown with a black line. Visual networks (VI), cerebellar networks (CB), temporal networks (TP), subcortical networks (SC), somatomotor networks (SM), and high cognitive processing networks (HCP). First-degree relatives with a diagnosis of a psychotic disorder were excluded from these analyses.