Preprint: https://www.medrxiv.org/content/10.1101/2024.03.05.24303772v1
Now published in Nature: https://www.nature.com/articles/s41586-024-08217-y
This repository contains code to perform the analyses for this project.
1_1_heritability_GCTA-LDMS.sh: calculates heritability of neurodevelopmental conditions (NDCs) using LD score regression.
1_2_heritability_LDSC_meta_GWAS.sh: calculates genetic correlations between NDCs and brain-related traits and conditions using LD score regression.
1_3_rgen_external_traits_with_NDC_conditioning_on_EA_GenomicSEM_mediation.R: calculates genetic correlations conditioning on educational attainment (EA) using GenomicSEM and the proportion of the genetic correlation explained by the latent genetic component of NDCs that is correlated with EA.
2_1_standardisation_PGS_t_test_comparing_subgroups.R: standardises polygenic scores (PGS) using control individuals, calculates average PGS in subgroups, and performs two-sample t-tests to compare subgroups.
2_2_mean_PGS_MCS_weighting.R: applies statistical weights adjusting for sampling and non-response bias in MCS to calculate weighted average PGS that are likely to be representative of the full UK population.
3_PGS_genetic_diagnosis_factors.R: performs regression models to test for associations between PGSs and diagnostic status, and whether associations between the chance of getting a diagnosis and factors affecting it might be confounding, or be confounded by PGS.
4_pTDT_analysis.R: performs pTDT to assess if unaffected parents over-transmit polygenic risk to undiagnosed probands.
5_trio_model_non_transmitted_coefficient.R: runs the proband-only model and the trio model, with the latter assessing whether parental alleles not transmitted to the child are correlated with the child's risk of NDCs.
6_1_correlation_RVBS_PGS_NDC.R: calculates the correlation between the rare variant burden score (RVBS) in one parent and the PGS in the other parent, as well as the correlation within the same parent or in the same proband.
6_2_correlation_RVBS_PGS_MCS_weighted.R: calculates similar correlations in a control cohort, with and without applying the weights to correct for sampling and non-response bias in MCS.