Single-cell RNA-clocks

Saket Choudhary

saketc@iitb.ac.in

Computational Multi-omics of Ageing

DH 603

Lecture 09 || Wednesday, 9^ththApril 2025

dfdf

Can we predict age in single-cells using transcriptome alone?

Source

Buckley et al. 2022

dfdf

Cell cycle stages

Source

Source

Stages:

• G1 = Beginning to divide
• S = Make copies of DNA
• G2 = Organize and condense the genetic material
• M = Mitosis

dfdf

Training data from 4 independent cohorts

Source

scRNA-seq data of SVZ neurogenic regions from four independent cohorts of 4–8 male mice, aged 3.3 to 29 months

How is scRNA-seq typically processed?

6

Key steps of any single-cell technology

Dissociation

Physically separate cells

Cell lysis

Reverse transcription

Tissue of interest

Pooled sequencing

7

Single-cell analysis is a multi-step workflow

​

Leucken and Theis 2019

• Raw sequencing data are processed and aligned to give count matrices
• Count matrices tabulate how many unique copies of mRNA are sequenced for each gene and each cell (also called UMIs)
• The count data (UMI matrix) undergo pre‐processing and downstream analysis

8

Single-cell analysis workflow

​

scRNA-seq unique molecular counts (UMIs) matrix

Input of any scRNA-seq workflow:

Image credits:

Azenta.com

Question: How should we ‘normalize’ counts matrix to adjust for non-biological variation?

Counts matrix needs to be ‘normalized’ before any downstream analysis to account for difference in total molecules sequenced

1

1

dfdf

How does the method work?

Source

Goal: Generate a training dataset that equally weighted each sample.

​

• From each cell type and sample, 15 cells are sampled and combined to generate one BootstrapCell
• Repeated previous step 100 times per cell type and sample combination
• Run Lasso for feature selection

Biological age = Proliferative fraction (cells predicted to be G2/M or S phase)

​

​

dfdf

How does the method work?

Source

EnsemblCells:

• For each model, partition scRNA-seq data into groups of 15 single-cell transcriptomes
• Sum the unique transcript counts for all cells in each group to create ‘EnsembleCells’
• Predict age using the weighted average of predictions across all 20 models, where weights were determined by the coefficient of variation (R²) on held out validation set

​

​

dfdf

Performance of approaches across celltypes

Source

dfdf

Generalization to independent datasets

Source

dfdf

What genes underlie the ageing clocks?

Source

dfdf

What genes underlie the ageing clocks?

​

Source

Ageing clocks are highly celltype-specific

dfdf

Parabiosis

Source

Source

Mice are joined at the thoracic and abdominal area, skin-to-skin, for 8 weeks before analysis

Isochronic pairings = either young–young or aged–age

dfdf

Effect of heterochronic parabiosis on cell-type-specific aging clocks.

​

Source

dfdf

Effect of heterochronic parabiosis on cell-type-specific aging clocks.

Source

dfdf

Effect of exercise on cell-type-specific aging clocks.

Source

dfdf

Effect of exercise on cell-type-specific aging clocks.

Source

dfdf

Questions