jkh's picture
Add README.md
00968c6 verified
This data was collected from the Nature publication
"A single-cell transcriptomic atlas characterizes ageing tissues in the mouse"
DOI: https://doi.org/10.1038/s41586-020-2496-1
Gene Expression Omnibus (GEO) Accession Code: GSE132042
In this study, researchers looked at how much certain genes were expressed in
young and old mice so that they could compare gene expression between the groups.
This provides data on how gene expression changes with age. Specifically, the
samples were taken from male and female mice. The samples from 'young' mice came
from mice at 1 and 3 months of age, and the samples from 'old' mice came from mice
at 18, 21, 24, and 30 months of age.
Also, the researchers used three different methods to measure gene expression:
1. FACS
2. Droplet
3. Bulk
This redundancy helps cross-validate the data. FACS and droplet are both single-cell
techniques, which means that gene expression is measured from each cell in a sample
individually. In contrast, the bulk approach examines many cells together (in bulk)
and basically finds their average expression rates. The metrics are reported by
technique, and the researchers also reported the four metrics on the data combined
from all three techniques.
The methods provided four different metrics for gene expression:
1. raw_p: the p-value, a measure of significance that the difference in expression
between the young and old groups is actually different
2. bh_p: the Benjamani-Hochberg p-value. This is a p-value adjusted for the huge
number of hypotheses (genes) being tested at once. This is useful because by chance,
some of the expression differences would appear random, and the BH p-value accounts
for this
3. coef: the age coefficient. If you were to fit a line of best fit to how the
gene expression rate changes with age, basically this would be the slope of the line.
So if a gene increases in expression with age, the age coefficient is positive, and
if a gene decreases in expression with age, the age coefficient would be negative.
4. fc: fold change. This quantifies how much the gene expression level differs
between the young and old mice. P-value and BH p-value indicate *whether* there's
a difference, and this metric quantifies the strength of the difference.
The researchers also looked at how gene expression rates differed by tissue. This is
useful because different tissues express genes at different rates. For example,
tissue X might express gene G a *lot*, which tissue Y might express it very little
or not at all. The tissues sampled were:
- Aorta
- BAT (brown adipose tissue)
- Bladder
- Brain myeloid
- Brain non-myeloid
- Diaphragm
- GAT (gonadal adipose tissue)
- Heart
- Kidney
- Large intestine
- Limb muscle
- Liver
- Lung
- MAT (marrow adipose tissue)
- Mammary gland
- Marrow
- Pancreas
- SCAT (SubCutaneous Adipose Tissue)
- Skin
- Small intestine
- Spleen
- Thymus
- Tongue
- Trachea
Note that not every method was applied to each tissue type.
The four metrics are also reported by cell type. For example, BAT.T Cell means
that the sample was taken from T cells in BAT tissue.
The comparison_summary table shows which methods provided data on which genes.
For example, the cell in the FACS column and the Aorta row says 1001, and you'll
see in the FACS_Aorta file that the four metrics are provided for 1001 genes.