layout	default
title	Cross-species analysis of Alzheimer’s pathology

Cross-species analysis of Alzheimer’s pathology

ADA 2025 — EPFL
Team: Sharknut
Project focus: Assessing the relevance of the TgCRND8 mouse model for Alzheimer’s disease using Xenium spatial transcriptomics.

---

Motivation

Alzheimer’s disease (AD) is a complex neurodegenerative disorder whose molecular mechanisms remain only partially understood. Mouse models are widely used to study AD pathology, yet their relevance for human disease—particularly at the level of cell-type-specific transcriptional programs—remains an open question.

Recent advances in spatial transcriptomics, such as 10x Genomics Xenium, enable high-resolution, in situ profiling of gene expression across intact tissue sections. This provides a unique opportunity to directly compare disease-associated transcriptional changes across species within a shared anatomical context.

In this project, we investigate to what extent the TgCRND8 mouse model recapitulates human AD-related transcriptional alterations in the cortex.

---

Data

We analyze two Xenium spatial transcriptomics datasets:

• Mouse: TgCRND8 (AD model) vs WT mice at multiple ages
• Human: Alzheimer’s disease vs healthy control cortical samples

For each dataset, we use:

• Cell segmentation and cell-type annotations
• A predefined gene panel shared across samples
• Spatially resolved gene expression at single-cell resolution

Only cortical cells are retained for downstream analyses.

---

Research questions

• RQ1: How do cell-type compositions differ between disease and control conditions across species?
• RQ2: To what extent are disease-associated transcriptional programs conserved between mouse and human cortical cell types?

This website focuses primarily on RQ2.

---

Methods overview

At a high level, our analysis pipeline includes:

• Quality control and gene filtering based on expression prevalence
• Cell-class stratification into major neuronal and glial populations
• Differential expression analysis within each species and cell class
• Pathway enrichment analysis of up- and down-regulated genes
• Cross-species comparison of transcriptional signals at gene and pathway levels

Detailed implementation is available in the project repository.

---

RQ2 — Cross-species transcriptional convergence

What does “convergence” mean?

We define cross-species convergence as the presence of transcriptional changes that are:

• Associated with disease status in both species
• Observed within the same broad cell class
• Directionally consistent (up- or down-regulated)

We assess convergence at both the gene level and the pathway level.

---

RQ2.1 — Gene filtering and differential expression context

Before comparing transcriptional signals across species, it is essential to contextualize:

• How many genes pass expression filters in each species
• How many of these genes are identified as differentially expressed (DE)

This step determines the depth and sensitivity of downstream pathway analyses.

Figure X: Fraction of genes retained after filtering and number of differentially expressed genes across species and cell classes.

Key observation:
(1–2 sentences summarizing the main pattern you observe here.)

---

RQ2.2 — Pathway-level overlap across species

To move beyond individual genes, we perform pathway enrichment analysis separately in mouse and human for each cell class.

We then quantify cross-species overlap between enriched pathways, accounting for differences in enrichment depth across species.

Figure X: Cross-species pathway overlap across cell classes.

Key observation:
(1–2 sentences summarizing glial vs neuronal patterns, or convergence vs divergence.)

---

RQ2.3 — Concordant genes and Alzheimer’s relevance

At the gene level, we examine which differentially expressed genes are:

• Shared between species
• Directionally concordant
• Associated with Alzheimer’s disease

Particular attention is given to genes with known AD relevance.

Figure X: Gene-level concordance between mouse and human.

Key observation:
(1–2 sentences highlighting scarcity of concordant genes and any standout examples.)

---

RQ2 wrap-up

Across cortical cell classes, we observe:

• Limited gene-level concordance between mouse and human
• Stronger alignment at the pathway level, particularly in glial populations
• Marked differences between neuronal and non-neuronal cell classes

Together, these results suggest that while the TgCRND8 model captures certain higher-level disease-associated processes, substantial species-specific regulation persists.

---

Limitations

Several limitations should be considered:

• The analysis is restricted to a predefined Xenium gene panel
• Differential expression yield differs markedly between species
• Human and mouse samples differ in age structure and disease progression
• Cell-type annotations may not be perfectly aligned across species

---

Key takeaways

• Cross-species transcriptional convergence in AD is limited at the gene level
• Pathway-level signals reveal partial conservation, especially in glial cells
• Neuronal populations show weaker cross-species alignment
• The TgCRND8 model captures aspects of AD biology but does not fully recapitulate human disease