Causal inference in integrative genomics: exploring de novo mutations and unravelling causal mechanisms via gene regulatory networks in developmental process

Integrative genomics has revolutionised our understanding of complex biological processes by synergistically combining data from various omics technologies, including genomics, transcriptomics, and epigenomics. In this study, we first focused on integrating genomics and transcriptomics data to perform causal inference on DNMs occurring in the germline. These germline DNMs have long been implicated as significant contributors to developmental and genetic disorders. Previous investigations of the genetic factors influencing DNMs have been hindered by limited statistical power, primarily due to the challenge of obtaining an adequate number of parent-offspring trios. However, leveraging the rare disease cohort from the UK's 100,000 Genomes Project (100kGP), which comprises over 10,000 trios, provided an unprecedented opportunity to delve into the genetics of germline mutations. Here, we estimated Single Nucleotide Polymorphism (SNP) heritability of DNM count in offspring, which measures the relative contribution of genetic factors to the variance of the trait, in a Principal Component Analysis (PCA)-selected subset of the 100kGP cohort. To gain deeper insights, we separately estimated SNP heritabilities for paternally and maternally transmitted mutations, determined based on parentally phased DNMs in offspring. This estimation was performed using parental genetic variants at various minor allele frequencies and through employing diverse methodologies. Here, although we were not able to find statistically significant evidence for the non-zero heritability of the DNM counts, we were able to explore how the heritability might be sensitive to the structure of the population and to confirm the increased significance of environmental factors in the context of causality of DNM. We further investigated the partitioning of heritability among genes with different expression profiles in different tissues or cells and found a relative heritability enrichment for genes expressed in gonadal tissues, particularly testis. Furthermore, we applied Mendelian randomisation approach with phenome-wide association studies to identify environmental factors affecting the generation of DNMs together with UK Biobank databases. We ended up with the conclusion that early onset of menopause is associated with an increased load of DNMs. As a study distinct from the one mentioned earlier, by integrating transcriptomics data and network information based on protein-protein interactions, we inferred gene regulatory networks (GRNs) operating during developmental processes. To achieve this, we utilised single-cell RNA-seq data from node-ablated and normal chick embryos during gastrulation and conducted differential expression and trajectory analyses, and then identified differentially expressed transcription factors (TFs) and determined the pseudo time for the formation of a specific developmental pathway, such as mesodermal and early endodermal pathway. Based on expression level data labelled by this pseudo time, we explored the characteristics of the GRNs inferred by a deep graph neural network in both normal and node-ablated embryos, aiming to identify potential topological or transcriptional differences between the two networks. By integrating diverse omics data and employing advanced computational approaches, this study provides valuable insights into the causality of genetic and environmental basis of DNMs and sheds light on the regulatory mechanisms governing complex biological systems during development.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2024

Author dc:creator

• Hwang, Seongwon

Advisor dc:contributor.advisor

• Scally, Aylwyn

• whole genome sequencing
• de novo mutation
• statistical genetics
• causal inference
• graph neural network
• 100K genomes project
• heritability estimation
• Mendelian randomization
• node ablation

Licence

• https://www.repository.cam.ac.uk/bitstreams/0ca37725-6fb6-49f7-ab91-e31bb6448f78/download
• http://purl.org/NET/rdflicense/allrightsreserved

Language dc:language

eng

DOI dc:identifier.doi

https://doi.org/10.17863/CAM.115039

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/378760