Medaka fish as a model for complex trait genetics

In this thesis, I use the medaka fish (Oryzias latipes) and the Medaka Inbred Kiyosu-Karlsruhe (MIKK) panel — a genetic resource established by my supervisor, Prof. Ewan Birney, and collaborators — to investigate the genetic architecture of complex traits, with a particular focus on non-additive genetic effects. My research addresses two phenotypes: one physiological (heart rate) and one behavioural (bold-shy behaviour). After providing an overview of the history of genetics and the current state of the art in Chapter 1, the main body of the thesis begins in Chapter 2 with a description of the computational methods I developed during my doctoral work. These methods underpin the analyses presented in the subsequent chapters and are also described in dedicated publications (Pierotti, Fitzgerald & Birney, 2025; Pierotti et al., 2024). They include genotype imputation from shallow sequencing data, mixed-model Genome-Wide Association Studies (GWAS), and post-GWAS causal variant prioritisation. In Chapter 3, I investigate heart rate in medaka embryos— a phenotype relevant to human physiology— and its variation across inbred strains exposed to different temperatures. In collaboration with Dr. Bettina Welz (BW, Wittbrodt group, Heidelberg University), I found that both baseline heart rate and its thermal response are highly heritable within the MIKK panel. Following this initial screen, we selected eight strains with contrasting heart rates and divergent responses to temperature, from which we generated a 10-way F2 cross comprising over 2000 embryos. Linear mixed model-based GWAS in this population identified 16 genome-wide significant loci associated with heart rate or thermal response. I pinpointed likely causal genes for four of these loci, which BW subsequently validated through genome editing. Notably, two of the putative causal variants discovered in the F2 cross were also detected in wild medaka from Japan, confirming their relevance in natural populations. The 16 loci described in Chapter 3 exhibited widespread non-additive genetic effects, including dominance, gene-by-environment (G×E) interactions, epistasis (G×G), and gene-by-gene-by- environment (G×G×E) effects. In Chapter 4, I detail the statistical evaluation of these effects, demonstrating their substantial contribution to phenotypic variance in a vertebrate population derived from wild haplotypes. To assess the broader implications for human complex trait genetics, I constructed a simulation framework to evaluate how statistical model choice influences QTL detection in the presence of non-linear effects. These simulations show that non-additive effects of the magnitude observed in medaka often fall below the detection threshold of current human association studies, offering an explanation for the apparent additivity observed in human trait architectures assessed by outbred GWAS despite widespread non-linearity in model organisms. v This medaka heart rate study and the accompanying statistical analysis are available in my co-first author preprint (Welz et al., 2025), which is currently under review for publication. In Chapter 5, I investigate genetic and social genetic effects on behavioural variation in medaka. I focused on the bold-shy behavioural axis using classical Japanese inbred strains. Analysing video recordings from 307 fish pairs collected by Dr. Ian Brettell (IB, Birney group, EMBL-EBI), I applied a neural network-based tracking system and a Hidden Markov Model to infer latent behavioural states from movement patterns. I identified significant strain-specific behavioural differences, establishing a foundation for future genetic dissection using the MIKK panel. I recently published this analysis in a co-first author paper with IB (Pierotti, Brettell et al., 2025). Collectively, this thesis establishes medaka as a powerful model for the study of complex trait genetics and provides insights with direct relevance to human biomedical research. By combining controlled genetic crosses with statistical modelling, I demonstrate that non-additive genetic effects play a substantial role in phenotypic variation — effects that are often missed in human studies due to limited statistical power. The successful validation of causal variants through genome editing, and their persistence in wild populations, bridges the gap between laboratory findings and natural genetic diversity. Finally, the open-source computational pipelines developed during this work enable broader adoption of these methods, facilitating future discoveries in complex trait genetics across diverse model systems.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2025

Author dc:creator

• Pierotti, Saul

Advisor dc:contributor.advisor

• Birney, Ewan

• Complex traits
• Dominance
• Epistasis
• Genetics
• GWAS
• GxE
• Medaka

Licence

• https://www.repository.cam.ac.uk/bitstreams/df10aa9b-8e38-4812-b0f6-13661ea3c569/download
• https://creativecommons.org/licenses/by/4.0/

Language dc:language

eng

Author Identifier

0000-0002-5942-0793

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/393276