Project Summary Reproducible assembly of protein molecules from free amino acids requires external information stored in DNA/RNA (genome). In contrast, assembly of higher-order structures from individual protein molecules is thought to occur de novo, relying on information intrinsic to each protein component. How higher-order complexity emerges efficiently without error using only the scattered information in each building block remains a question hindering core understandings of life. Here we address this question using the biogenesis of the centriole organelle as the example. We found that a proteinaceous structure called the cartwheel is grown inside existing centrioles, which is then split, shed and used as structural templates (or guidance) for reproducing a copy of the centriole with the same size/shape. Uncontrolled cartwheel shedding was further found to underly a human disease called Alström syndrome. By erasing the old guidance, we showed that new/naive centrioles varying widely in size and shape arise, which can propagate reproducibly thereafter in cells otherwise genetically identical. The simplest model explaining our results is that one or more steps of centriole biogenesis are template-based, instructed by information stored and transmitted outside of the genome. This model on which the current proposal is based on will shed light on how cellular complexity emerges reproducibly from components randomly distributed in the cell, redefining the scope of biological information as we currently understood.