Kavli Affiliate: Joshua Vogelstein
| Authors: Jaewon Chung, Eric W Bridgeford, Michael Powell and Joshua T Vogelstein
| Summary:
The heritability of human connectomes is crucial for understanding the genetic and environmental factors that influence variations in connectomes, which can provide insights into behavior and disease. However, current approaches to studying connectome heritability assume an associational effect and often rely on modeling assumptions that may not hold true for complex, high-dimensional connectome data. In this study, we propose a causal perspective to investigate connectome heritability, using statistical models that capture the underlying structure and dependence within connectomes. These models allow us to explicitly define different notions of connectomic heritability by removing common structures with increasing complexity across connectomes. We develop a non-parametric test to detect and test for these notions of heritability and apply it to connectomes estimated from the Human Connectome Project (HCP) diffusion data, demonstrating their heritability after accounting for confounding variables such as brain anatomy, age, and sex. This work highlights the potential of using statistical modeling of networks and causal methods to study connectome heritability, ultimately providing a better understanding of the genomic influences on brain connectivity.