Extended Data Fig. 1: Workflow for the generation of human CA1 circuit.

Gray boxes represent the adopted procedural blocks. Red boxes represent computational functions. Green box represents non elaborated data obtained from public repositories (external source). 1 Neuronal placement: Cells coordinates have been downloaded from the BigBrain image database in the form of grayscale images. Images have been segmented and hippocampus subregions have been isolated (CA1, CA2, CA3, CA4, Subiculum). Putative cell soma positions have been isolated through image analysis. Subsequently, a pruning procedure allowed to assign the 3D coordinates to excitatory (1) and inhibitory (7) classes according to their relative positions in CA1 layers (OUT1a). CA2 and Subiculum surface meshes have been generated (OUT1b). 2 Neuronal morphologies: Geometrical shapes were generated by assigning morphological parameters to dendritic and axonal features. Neurons were oriented according to external landmarks and axonal convex hulls (OUT2a) and dendritic point volumes (OUT2b) were generated. 3 Network connectivity; Bounding boxes for axons and dendrites were generated and an iterative algorithm calculated the intersections of axonal hull with dendrites showing overlapping bounding boxes. A connection pair (OUT3) was generated when at least one dendritic point fell into the axonal hull. 4 Network simulation. The scaffold and connectivity matrix were loaded into NEST simulator to simulate CA1 activity. The PMA algorithm is represented by blocks 2 and 3.