First, they’re associated with the ‘costs’ of network formation. Second, they define ‘high value’ areas within the network.
How she did it:
First, she built connectomes from ~2k 10yr olds from ABCD. To each she fit a generative network model (GNM) – a simple computational model for simulating their topology (Figure 1.)
Those with *high* polygenic scores (PGS) for cognitive ability had a *softer* ‘distance penalty’ in their GNM. The end result is that networks form slightly more ‘shortest paths’. The end result? Slightly more stochastic, variable and efficient brain networks (Figure 2.).
Second, she used transcription data from the @AllenInstitute, and found that the genes in the PGS particularly overlap with the regional distribution of the ‘values’ part of the GNM – that is they define the areas the model wants to connect (Figure 3.).
These genes in particular seem to be responsible for cellular and biological processes involved in synapse development and transmitter and anion-gated channel activity.
Want the details? Check out the new open access paper: Brain wiring economics, network organisation and population-level genomics
Acknowledgements go to funders: @MRC , @TempletonWorld and supporters: @CamNeuro, @MRCCBU, @Cambridge_Uni.
—
Adapted from X Thread by @DuncanAstle.
Posted on 30/07/2025
