interview prep

Why are you interested in this PhD?

• Focuses on vascular contributors to SVD and AD, which are often treated as confounders rather than modifiers

• Uses multi-omic integration (genetics, epigenetics, imaging, RNA-seq, proteomics, blood biomarkers)

• Epigenetics helps explain why shared genetic risk leads to different disease trajectories

• Studying SVD and AD together reflects clinical reality of co-pathology

Why epigenetics rather than genetics alone?

• Genetics is static; epigenetics is dynamic and context-dependent

• Many GWAS variants are non-coding and act through regulatory mechanisms

• Epigenetics explains when, where, and in which cell types risk is expressed

• Acts as an interface between genotype, environment, and phenotype

How do Alzheimer’s disease and SVD relate to each other?

• Biologically related but mechanistically distinct diseases

• Share ageing as an upstream risk factor and cognitive decline as a downstream outcome

• APOE is a major AD risk factor but not consistently linked to WMHs

• SVD acts as a disease modifier, not a direct cause or confounder

How does your pharmacology background prepare you for this PhD?

• Training in molecular mechanisms of monogenic and polygenic diseases

• Strong understanding of non-coding variants and regulatory effects

• Pharmacology emphasises modulation, dose–response, and intervenable pathways

• Aligns with epigenetics as a regulatory and modifiable layer

Association vs causation in GWAS and epigenetics

• GWAS identifies statistical associations, not causal mechanisms

• Loci often contain multiple variants in linkage disequilibrium

• Epigenetics provides biological context (cell type, timing, state)

• Neither proves causation alone, but together help prioritise mechanisms

How does Mendelian randomisation help?

• Uses genetic variants as instrumental variables

• Tests whether a genetically predicted exposure is consistent with causality

• Steiger filtering assesses direction of effect

• MR-Egger helps detect horizontal pleiotropy

• Stronger when combined with epigenetic biological context

What are the main limitations of epigenetic studies?

• Difficult to disentangle cause vs consequence

• Strong cell-type heterogeneity and bulk tissue bias

• Highly context-dependent (age, environment, disease stage)

• Requires longitudinal and integrative study designs

Why VIDA DTC specifically?

• Treats vascular biology as central, not secondary

• Emphasises causal inference and multi-omic integration

• Focus on regulation rather than single-gene causation

• Strong alignment with co-pathology and systems-level thinking

What are the key challenges in studying vascular dementia?

• Vascular processes are dynamic, age-dependent, and physiological

• Genetic effects are small and diffuse, making detection difficult

• Imaging markers (e.g. WMHs) reflect accumulated damage, not active pathology

• Requires longitudinal approaches to separate drivers from consequences

What characterises a good biomarker?

• Disease-relevant and detectable early

• Trackable longitudinally

• Balances usefulness vs specificity

• Ideally non-invasive and multimodal

Why blood-based data for a brain disease?

• Captures systemic vascular, immune, and metabolic processes

• Suitable for longitudinal and scalable sampling

• Reflects upstream drivers of brain vulnerability

• Most powerful when integrated with imaging and brain measures

What would you do if your hypothesis was wrong?

• Negative results are informative, not failures

• Reassess assumptions and refine models

• Use data to constrain alternative hypotheses

• Complexity means insight often comes from unexpected findings

How do you handle uncertainty in research?

• Accept uncertainty as inherent to complex biology

• Triangulate evidence across methods

• State assumptions clearly and avoid over-interpretation

• Value convergent evidence over single results

What skills do you want to develop?

• Causal inference and multi-omic integration

• Independent question framing and hypothesis testing

• Translating complex data into biological insight

• Developing intellectual independence

What would success look like after the PhD?

• Robust, reproducible findings

• Clear mechanistic understanding

• Intellectual independence

• Publications are important, but insight matters more

Why should we choose you?

• Strong conceptual and methodological fit

• Comfortable with complexity and co-pathology

• Causally disciplined, integrative thinker

• Aligned with the programme’s systems-level aims

Why do many GWAS hits fail to translate into mechanisms?

• GWAS identifies loci, not causal variants or genes

• Most hits are non-coding and act through regulation

• Linkage disequilibrium obscures true causal variants

• Functional context (cell type, timing) is missing without epigenetics

How does epigenetics add value beyond transcriptomics?

• Transcriptomics captures gene expression state, not regulation

• Epigenetics reveals why and how expression is permitted or restricted

• Regulatory changes may precede detectable expression changes

• Epigenetics provides cell-type–specific control mechanisms

What are the key assumptions behind Mendelian randomisation?

• Genetic variants act as instrumental variables

• Variants influence outcome only through the exposure

• No confounding of genotype–outcome relationship

• Horizontal pleiotropy must be assessed (e.g. MR-Egger)

When does Mendelian randomisation fail?

• Weak instruments reduce power

• Horizontal pleiotropy violates assumptions

• Complex traits with feedback loops complicate interpretation

• MR infers plausibility, not definitive causality

Why is small vessel disease difficult to model experimentally?

• Human SVD reflects chronic, lifelong processes

• Animal models poorly capture age-related vascular pathology

• Multiple interacting pathways (vascular, immune, metabolic)

• Disease manifests as diffuse network dysfunction, not focal lesions

Why are white matter hyperintensities difficult to interpret?

• Reflect accumulated damage, not active pathology

• Can arise from multiple upstream mechanisms

• Poor temporal resolution in cross-sectional studies

• Clinical impact varies by location and burden

How do vascular factors modify neurodegenerative disease?

• Alter brain resilience and vulnerability

• Lower threshold for clinical symptom expression

• Interact with protein pathology and inflammation

• Influence disease trajectory rather than initiation

What does “co-pathology is the rule, not the exception” mean?

• Multiple pathologies often coexist in ageing brains

• Pure single-disease states are uncommon

• Clinical symptoms reflect combined pathological burden

• Studying diseases in isolation limits explanatory power

What are the challenges of multi-omic integration?

• Different data types have different scales and noise

• Requires careful alignment across samples and time

• Risk of false convergence without biological grounding

• Interpretation is more challenging than single-modality analysis

Why is integration better than focusing on one “best” modality?

• No single modality captures disease complexity

• Genetics gives risk, epigenetics gives regulation, imaging gives phenotype

• Convergent evidence increases confidence

• Integration reduces over-interpretation of isolated findings

Why are vascular biomarkers clinically attractive?

• Vascular factors are modifiable

• Often detectable earlier than neurodegeneration

• Relevant across multiple dementia subtypes

• Align with prevention and risk stratification strategies

What are the limitations of blood-based biomarkers?

• Peripheral signals may not reflect brain-specific processes

• Influenced by systemic illness and lifestyle factors

• Require careful validation and integration

• Best used as risk or modifier markers, not sole diagnostics

What motivates you about complex, non-linear diseases?

• Reflect real biological systems

• Require integrative and critical thinking

• Negative or null results are informative

• Allow for hypothesis refinement rather than binary answers

How do you prioritise research questions?

• Focus on biological plausibility and tractability

• Prefer questions that integrate across scales

• Value relevance to disease heterogeneity

• Avoid overly narrow or purely descriptive aims

How would you approach a new dataset you don’t understand?

• Start with exploratory analysis and quality control

• Understand assumptions behind data generation

• Look for consistency across related variables

• Integrate with existing biological knowledge

Isn’t epigenetics just correlation?

• It can be, but it also reflects regulatory mechanisms

• Causal claims require integration with genetics and experiments

• Epigenetics helps prioritise plausible mechanisms

• Valuable even when not strictly causal

Why not focus on a single disease to reduce complexity?

• Complexity reflects clinical reality

• Co-pathology explains heterogeneity in outcomes

• Shared mechanisms emerge only through comparison

• Studying in parallel avoids false causal assumptions

What excites you most about this field right now?

• Shift toward systems-level understanding

• Recognition of vascular and immune modifiers

• Advances in multi-omic technologies

• Movement away from single-cause disease models