schoof2021.RdSingle-cell proteomics data from OCI-AML8227 cell culture to
reconstruct the cellular hierarchy. The data were acquired using
TMTpro multiplexing. The samples contain either no cells,
single cells, 10 cells (reference channel) 200 cells (booster
channel) or are simply empty wells. Single cells are expected to
be one of progenitor cells (PROG), leukaemia stem cells (LSC),
CD38- blast cells (BLAST CD38-) or CD38+ blast cells
(BLAST CD38+). Booster are either a known 1:1:1 mix of cells
(PROG, LSC and BLAST) or are isolated directly from the bulk
sample. Samples were isolated and annotated using flow cytometry.
schoof2021A QFeatures object with 194 assays, each assay being a SingleCellExperiment object:
F*: 192 assays containing PSM quantification data for 16
TMT channels. The quantification data contain signal to noise
ratios as computed by Proteome Discoverer.
proteins: quantitative data for 2898 protein groups in 3072
samples (all runs combined). The quantification data contain
signal to noise ratios as computed by Proteome Discoverer.
logNormProteins: quantitative data for 2723 protein groups in
2025 single-cell samples. This assay is the protein datasets that
was processed by the authors. Dimension reduction and clustering
data are also available in the reducedDims and colData slots,
respectively
Sample annotation is stored in colData(schoof2021()). The cell
type annotation is stored in the Population column. The flow
cytometry data is also available: FSC-A, FSC-H, FSC-W, SSC-A,
SSC-H, SSC-W, APC-Cy7-A (= CD34) and PE-A (= CD38).
The PSM and protein data can be downloaded from the PRIDE repository PXD020586 The source link is: https://www.ebi.ac.uk/pride/archive/projects/PXD020586
The data were acquired using the following setup. More information
can be found in the source article (see References).
Sample isolation: cultured AML 8227 cells were stained with anti-CD34 and anti-CD38. The sorting was performed by FACSAria instrument and deposited in 384 well plates.
Sample preparation: cells are lysed using freeze-boil and sonication in a lysis buffer (TFE) that also includes reduction and alkylation reagents (TCEP and CAA), followed by trypsin (protein) and benzonase (DNA) digestion, TMT-16 labeling and quenching, desalting using SOLAµ C18 plate, peptide concentration, pooling and peptide concentration again. The booster channel contains 200 cell equivalents.
Liquid chromatography: peptides are separated using a C18 reverse-phase column (50cm x 75 µm i.d., Thermo EasySpray) combined to a Thermo EasyLC 1200 for 160 minute gradient with a flowrate of 100nl/min.
Mass spectrometry: FAIMSPro interface is used. MS1 setup: resolution 60.000, AGC target of 300%, accumulation of 50ms. MS2 setup: resolution 45.000, AGC target of 150, 300 or 500%, accumulation of 150, 300, 500, or 1000ms.
Raw data processing: Proteome Discoverer 2.4 + Sequest spectral search engine and validation with Percolator
All data were collected from the PRIDE repository (accession ID:
PXD020586). The data and metadata were extracted from the
SCeptre_FINAL.zip file.
We performed extensive data wrangling to combine al the metadata
available from different files into a single table available using
colData(schoof2021).
The PSM data were found in the bulk_PSMs.txt file. Contaminants
were defined based on the protein accessions listed in
contaminant.txt. The data were converted to a QFeatures
object using the scp::readSCP() function.
The protein data were found in the bulk_Proteins.txt file.
Contaminants were defined based on the protein accessions listed
in contaminant.txt.The column names holding the quantitative
data were adapted to match the sample names in the QFeatures
object. Unnecessary feature annotations (such as in which assay
a protein is found) were removed. Feature names were created
following the procedure in SCeptre: features names are the
protein symbol (or accession if missing) and if duplicated
symbols are present (protein isoforms), they are made unique by
appending the protein accession. Contaminants were defined based
on the protein accessions listed in contaminant.txt. The data
were then converted to a SingleCellExperiment object and
inserted in the QFeatures object.
The log-normalized protein data were found in the bulk.h5ad file.
This dataset was generated by the authors by running the notebook
called bulk.ipynb. The bulk.h5ad was loaded as an AnnData
object using the scanpy Python module. The object was then
converted to a SingleCellExperiment object using the
zellkonverter package. The column names holding the quantitative
data were adapted to match the sample names in the QFeatures
object. The data were then inserted in the QFeatures object.
The script to reproduce the QFeatures object is available at
system.file("scripts", "make-data_schoof2021.R", package = "scpdata")
Schoof, Erwin M., Benjamin Furtwängler, Nil Üresin, Nicolas Rapin, Simonas Savickas, Coline Gentil, Eric Lechman, Ulrich auf Dem Keller, John E. Dick, and Bo T. Porse. 2021. “Quantitative Single-Cell Proteomics as a Tool to Characterize Cellular Hierarchies.” Nature Communications 12 (1): 745679. (link to article).
# \donttest{
schoof2021()
#> see ?scpdata and browseVignettes('scpdata') for documentation
#> loading from cache
#> An instance of class QFeatures containing 194 assays:
#> [1] F1: SingleCellExperiment with 4455 rows and 16 columns
#> [2] F10: SingleCellExperiment with 4604 rows and 16 columns
#> [3] F100: SingleCellExperiment with 5056 rows and 16 columns
#> ...
#> [192] F99: SingleCellExperiment with 3898 rows and 16 columns
#> [193] proteins: SingleCellExperiment with 2898 rows and 3072 columns
#> [194] logNormProteins: SingleCellExperiment with 2723 rows and 2025 columns
# }