Periodic Reporting for period 3 - GEPSI (Genes, Policy, and Social Inequality)
Berichtszeitraum: 2024-01-01 bis 2025-06-30
Many important indicators of social status (such as one’s level of education, occupation, and income) have been shown to be moderately heritable, meaning that a part of the variation in social status can be explained by genetic differences across population members. By methodologically advancing the estimation of the interaction between genes and environments, this project will settle two long lasting debates in social science genetics.
First, this project will show how heritability studies –despite earlier firm rejections of this position– can be informative for policies aiming to reduce social inequalities. Second, the project team will assess the critique that social science genetics attributes effects to genes which should be attributed to the environments through which these genes operate. In doing so, we will extend existing methodology to quantify the incremental explanatory power of genes over environmental factors such as the social status of parents, and we will develop a solution for the endogenous inclusion of environmental factors in genetic summary indices (i.e. polygenic risk scores) which currently impacts the validity of gene-by-environment studies.
All newly developed methods will be tested extensively using simulations, and made available for others by means of free software code. Empirically, we will interact genes and various natural experiments (policy changes) to identify interventions that can ameliorate social inequalities in terms of education, occupational status, and income.
First, this project will show how heritability studies –despite earlier firm rejections of this position– can be informative for policies aiming to reduce social inequalities. Second, the project team will assess the critique that social science genetics attributes effects to genes which should be attributed to the environments through which these genes operate. In doing so, we will extend existing methodology to quantify the incremental explanatory power of genes over environmental factors such as the social status of parents, and we will develop a solution for the endogenous inclusion of environmental factors in genetic summary indices (i.e. polygenic risk scores) which currently impacts the validity of gene-by-environment studies.
All newly developed methods will be tested extensively using simulations, and made available for others by means of free software code. Empirically, we will interact genes and various natural experiments (policy changes) to identify interventions that can ameliorate social inequalities in terms of education, occupational status, and income.
The Genes, Policy, and Social Inequality (GEPSI) project is organized around three Work Packages (WP1, 2A, 2B) for which three researchers were hired. The Principal Investigator (PI) of the project is involved in each work package and works together with the recruited staff to realize the objectives of the project.
For WP 1, a PhD student was hired. Work unit 1 of WP1 (“Linking heritability and inequality of opportunity”) has been finished and is forthcoming as a chapter in the Research Handbook on Intergenerational Mobility (publication of the handbook is expected at the end of calendar year 2023). Work unit 2 and 3 have an empirical focus and draw on data from Statistics Netherland. We obtained access to these data, and we are currently finalizing the empirical analyses for two separate papers. The first paper addresses interrelationships among estimates of intergenerational transmissions (including heritability estimates) for educational attainment in the Netherlands. The second paper investigates gene-environment interplay in educational attainment.
For WP2A, a PostDoc was hired. So far, we completed work unit 1 (publications in Communications Biology and BMC Bioinformatics, and the software tool MGREML being available on GitHub) as well as work unit 2 (publication in PLOS Genetics, including further development of the software tool MGREML on GitHub). We started working on work unit 3 (“Development of Multivariate GREML for G×E analysis”), and the challenge we currently face is getting the optimization procedure finetuned for this type of modelling.
For WP2B, another PostDoc was hired. We published two papers about problems concerning polygenic scores, one about the rank (dis)concordance of polygenic scores (Nature Human Behaviour) and one about how to correct for measurement error in polygenic scores (Nature Communications). Using the methodology developed in the latter study we analyzed dynamic complementarity in skill production using UK Biobank data (working paper available on arXiv). These studies contribute to the realization of the overall objectives of the project because they all concern the validity of G×E studies using polygenic scores (Objective B) and they empirically analyze how the interaction between genes and environmental factors can ameliorate social inequalities.
The PI of this project on G×E interactions also contributed to a forthcoming (but already online available) publication in the Journal of Human Resources about the interplay between maternal smoking and genes using (amongst others) UK Biobank data, a review chapter about gene-environment interplay in the social sciences (published in the Oxford Research Encyclopedia of Economics and Finance), and a working paper about the economics and econometrics of gene-environment interplay (available on arXiv, R&R at Review of Economics Studies). These studies also contribute to the realization of the overall objectives of the project.
To foster the successful completion of the project, to transfer knowledge as well as to disseminate research findings, the PI co-organized the European Social Science Genetics Network Conference II (https://www.iser.essex.ac.uk/essgn2023(öffnet in neuem Fenster)). This interdisciplinary conference aimed to improve understanding of how genetic endowments and environmental circumstances shape life choices and outcomes across the life-course of individuals. Keynote speakers were Professor Kathryn Paige Harden (University of Texas at Austin) and Professor Nicole Soranzo (Human Technopole).
For WP 1, a PhD student was hired. Work unit 1 of WP1 (“Linking heritability and inequality of opportunity”) has been finished and is forthcoming as a chapter in the Research Handbook on Intergenerational Mobility (publication of the handbook is expected at the end of calendar year 2023). Work unit 2 and 3 have an empirical focus and draw on data from Statistics Netherland. We obtained access to these data, and we are currently finalizing the empirical analyses for two separate papers. The first paper addresses interrelationships among estimates of intergenerational transmissions (including heritability estimates) for educational attainment in the Netherlands. The second paper investigates gene-environment interplay in educational attainment.
For WP2A, a PostDoc was hired. So far, we completed work unit 1 (publications in Communications Biology and BMC Bioinformatics, and the software tool MGREML being available on GitHub) as well as work unit 2 (publication in PLOS Genetics, including further development of the software tool MGREML on GitHub). We started working on work unit 3 (“Development of Multivariate GREML for G×E analysis”), and the challenge we currently face is getting the optimization procedure finetuned for this type of modelling.
For WP2B, another PostDoc was hired. We published two papers about problems concerning polygenic scores, one about the rank (dis)concordance of polygenic scores (Nature Human Behaviour) and one about how to correct for measurement error in polygenic scores (Nature Communications). Using the methodology developed in the latter study we analyzed dynamic complementarity in skill production using UK Biobank data (working paper available on arXiv). These studies contribute to the realization of the overall objectives of the project because they all concern the validity of G×E studies using polygenic scores (Objective B) and they empirically analyze how the interaction between genes and environmental factors can ameliorate social inequalities.
The PI of this project on G×E interactions also contributed to a forthcoming (but already online available) publication in the Journal of Human Resources about the interplay between maternal smoking and genes using (amongst others) UK Biobank data, a review chapter about gene-environment interplay in the social sciences (published in the Oxford Research Encyclopedia of Economics and Finance), and a working paper about the economics and econometrics of gene-environment interplay (available on arXiv, R&R at Review of Economics Studies). These studies also contribute to the realization of the overall objectives of the project.
To foster the successful completion of the project, to transfer knowledge as well as to disseminate research findings, the PI co-organized the European Social Science Genetics Network Conference II (https://www.iser.essex.ac.uk/essgn2023(öffnet in neuem Fenster)). This interdisciplinary conference aimed to improve understanding of how genetic endowments and environmental circumstances shape life choices and outcomes across the life-course of individuals. Keynote speakers were Professor Kathryn Paige Harden (University of Texas at Austin) and Professor Nicole Soranzo (Human Technopole).
For WP 1, we are currently working on two papers (see previous box). These two papers are expected to be ready for submission to a scientific journal end of calendar year 2023. Thereafter, the PhD student will continue working, per the project proposal, on other empirical applications (e.g. on other measures of social status than educational attainment) using data from Statistics Netherlands.
For WP2A, we developed the software tool MGREML. Because of the development of a state-of-the-art memory-efficient and fast optimization algorithm, the MGREML software package allows multivariate analyses several orders of magnitude larger and faster than previously existing software packages. We consider this to be a significant achievement, although the gain in statistical efficiency turned out to be smaller than expected. Currently, we are working on finetuning the optimization procedure for Multivariate GREML G×E analysis (work unit 3). We expect to need time until end of calendar year 2023 for this. Thereafter we, will work on empirical applications using the newly developed model (per the project proposal).
For WP2B, we completed a manuscript on the endogenous inclusion of environmental factors in polygenic scores covering most aspects of work unit 1 (statistical formulation of problem) as well as of work unit 2 (literature review) and work unit 3 (developing methodological solution). Unfortunately the manuscript was rejected for publication at a scientific journal. The comments received are such that we have to redesign the statistical framework quite drastically, which we are planning on doing in the period September 2023-February 2024.
Regarding knowledge dissemination, the GEPSI team is planning on (co-)organizing the European Social Science Genetics Network Conference at our home institution (Erasmus University Rotterdam) in 2024.
For WP2A, we developed the software tool MGREML. Because of the development of a state-of-the-art memory-efficient and fast optimization algorithm, the MGREML software package allows multivariate analyses several orders of magnitude larger and faster than previously existing software packages. We consider this to be a significant achievement, although the gain in statistical efficiency turned out to be smaller than expected. Currently, we are working on finetuning the optimization procedure for Multivariate GREML G×E analysis (work unit 3). We expect to need time until end of calendar year 2023 for this. Thereafter we, will work on empirical applications using the newly developed model (per the project proposal).
For WP2B, we completed a manuscript on the endogenous inclusion of environmental factors in polygenic scores covering most aspects of work unit 1 (statistical formulation of problem) as well as of work unit 2 (literature review) and work unit 3 (developing methodological solution). Unfortunately the manuscript was rejected for publication at a scientific journal. The comments received are such that we have to redesign the statistical framework quite drastically, which we are planning on doing in the period September 2023-February 2024.
Regarding knowledge dissemination, the GEPSI team is planning on (co-)organizing the European Social Science Genetics Network Conference at our home institution (Erasmus University Rotterdam) in 2024.