ECHO
Expression Compendium of Hearing Omics
An integrated resource of bulk and single cell transcriptomes of the mouse inner ear, spanning development, ageing, noise exposure and ototoxic injury.
436
bulk samples
100+
studies
1,110,168
single cells
3
conditions
4
developmental stages
Bulk transcriptome
436 samples · 21,787 genes · 50 studies
Tissue and cell preparations
Cochlea189
Stria vascularis42
Hair cells32
OHCs30
Supporting cells29
IHCs26
SGNs21
Auditory nerve21
5 further categories46
Condition
Normal370
Noise53
Ototoxicity13
Developmental stage
Adult197
Early postnatal110
Aging80
Embryonic49
Single cell atlas
50+ studies · 1,110,168 cells · 44,488 genes · 14 cell types
Cell types
Vascular-associated cells234,493
Supporting/prosensory-like epithelial cells171,297
Cochlear chondrocytes161,114
Spiral ligament fibrocytes156,911
Myelinating Schwann cells85,048
Stria vascularis intermediate cells83,238
Reissner's membrane cells59,305
Spiral ganglion neurons58,579
6 further categories100,183
Condition
Normal894,937
Ototoxicity119,490
Noise95,741
Time stage
Adult370,219
Early postnatal360,149
Aging190,636
Embryonic189,164
What you can do here
Expression browsing
Query any gene across cell types, conditions and developmental stages with box plots, violin plots, dot plots and heatmaps.
Signature scoring
Score a gene set of your choice in every sample and every cell, with or without a background correction, and compare the result across groups.
Differential expression
Compare any two groups of bulk samples with a Mann-Whitney or Welch test and inspect the result as a volcano plot and table.
Cell type markers
Rank the genes that distinguish each cell type and see how they behave across the rest of the atlas.
Composition analysis
Track how the cellular composition of the inner ear shifts between conditions and across the lifespan.
Open data
Every figure and table exports to CSV, and the complete atlas is downloadable as an h5ad file.
Expression by group
Summary statistics
Expression heatmap
Gene set
Score by group
Signature score heatmap
Score per sample
Principal component analysis
Sample-to-sample correlation
Two-group comparison
Volcano plot
Differentially expressed genes
Bulk data

Expression values are log-normalised. Download the currently selected genes together with the sample annotation, or the complete sample annotation table.

Selected gene expressionSample annotation
Gene expression
Dot plot
Expression heatmap
Expression distribution
Expression by group
Gene set
Score on the UMAP
Score by group
Signature score heatmap
Score summary
Cell counts
Cell type markers
Marker dot plot
Marker table
About the data
Bulk transcriptome

436 mouse inner ear samples compiled from 50 independent studies and quantified for 21,787 genes. Every sample is annotated with its tissue or cell preparation, developmental stage, experimental condition and study of origin.

Expression values are log-normalised, so a difference between two group means approximates a log fold change. The study of origin is available as both a filter and a grouping variable. In the future, we plan to gradually incorporate more datasets and explore better standardisation across data generated by different studies.

Single cell atlas

1,110,168 cells, annotated into 14 cell types across 3 experimental conditions and 4 developmental or ageing stages. Expression values are log-normalised and UMAP coordinates come from a harmony-integrated embedding.

Dot plots, heatmaps, marker rankings, composition analyses and group level signature scores are computed over all 1,110,168 cells. Interactive cell level views use a subset that is stratified by cell type, condition and time stage so that every group remains represented. In the future, we plan to gradually incorporate data from more species and single-cell multi-omics datasets.

Analyses available on this site
  • Expression — distribution of a gene across groups, as box, violin or dot plots, and multi-gene heatmaps of group means or individual samples.
  • Signature score — the mean log expression of a user supplied gene set. The background-corrected option subtracts the mean of an expression-matched set of control genes, which removes the offset caused by differences in overall detection.
  • Differential expression (bulk) — Mann-Whitney or Welch t-test between two groups of samples, with Benjamini-Hochberg control of the false discovery rate.
  • Marker genes (single cell) — genes ranked by the difference between their mean expression in one cell type and in all others.
  • Cell composition (single cell) — cell type proportions across conditions and stages.
Data downloads
  • Complete single cell atlas — h5ad, compressed with the gzip filter built into HDF5, readable by scanpy, anndata, Seurat and zellkonverter without additional plugins.
  • Cell subset — the data behind the interactive views, small enough to reanalyse on a laptop. obs.sampling_weight gives the number of atlas cells each subset cell represents.
  • Bulk expression and sample annotation — CSV, from the bulk download tab.
  • Analysis results — every table on the site exports to CSV.