This group of tracks shows data from A human cell atlas of fetal gene expression. This is a collection of
single cell and single nucleus combinatorial indexing-based RNA-seq data covering 4 million
cells from 15 organs obtained during mid-gestation. The cells were sequenced in
a highly multiplexed fashion and then clustered with annotations as described
in Cao et al., 2020.
The Fetal Cells subtrack contains the
data organized by cell type, with RNA signals from all cells of a given type pooled
and averaged into one bar for each cell type. The
Fetal Lineage subtrack shows
similar data, but with the cell types subdivided more finely and by organ. Additional
bar chart subtracks pool the cell by other characteristics such as by sex
(Fetal Sex), assay
(Fetal Donor ID), experiment
(Fetal Exp), organ
(Fetal Organ), and reverse transcription group
(Fetal RT Group).
The cell types are colored by which class they belong to according to the following table.
The coloring algorithm allows cells that show some blended characteristics to show blended
colors so there will be some color variation within a class. The colors will be purest in
the Fetal Cells subtrack, where the bars
represent relatively pure cell types. They can give an overview of the cell composition
within other categories in other subtracks as well.
Three-level single-cell combinatorial indexing (sci-RNAseq3) as described in
Cao et al., 2020 was used on 121 samples from 28 fetuses estimated 72
to 129 days post-conception. This included samples from 15 organs. and
resulted in RNA profiles for 4 million cells. The samples were flash-frozen for
majority of the experiments and then nuclei extracted for sequencing. Samples
from tissues from the kidney and digestive system were fixed after
disassociation to deactivate endogenous RNases and proteases.
The cell/gene matrix and cell-level metadata was downloaded from the
UCSC Cell Browser. The UCSC command line utility matrixClusterColumns,
matrixToBarChart, and bedToBigBed were used to transform these into a bar chart
format bigBed file that can be visualized. The coloring was done by defining
colors for the broad level cell classes and then using another UCSC utility,
hcaColorCells, to interpolate the colors across all cell types.
The UCSC utilities can be found on
our download server.
Thanks to the many authors who worked on producing and publishing this data set.
The data were integrated into the UCSC Genome Browser by Jim Kent and Brittney Wick
then reviewed by Jairo Navarro. The UCSC work was paid for by the Chan Zuckerberg Initiative.