Skip to contents

Single-cell proteomic data from peripheral blood mononuclear cells (PBMCs) collected, isolated using the cellenONE, prepared using the N2 nanoPOTS technology, acquired with a ThermoFisher Orbitrap Eclipse Tribrid and searchd with FragPipe. Single cells were multiplexed using TMTpro 32-plex reagents (although 35 TMT reagents/set were found in the data). Cell types are not defined prior to data acquisition, but the data set also contains samples for boosting peptide identification ("Bridge"), empty sample for QC ("Blank") and empty samples that isotopically adjacent to bridge samples ("Isotope"). The data set contains data at the peptide and protein level as provided by the authors, but also modelled protein data using the scplainer approach.

Format

A QFeatures object with 3 assays, each assay inheriting from the SummarizedExperiment class:

  • peptides: formatted data generated by FragPipe (abundance_peptide_None.tsv), containing 8,723 peptides for 2,485 samples (mostly single cells, see annotations below).

  • proteins: same as the peptide data, but for proteins (using FragPipe's abundance_protein_None.tsv table).

  • peptides_modelled: same as peptides, but after minimal data processing and scplainer data modelling (see below).

The colData(petrosier2026()) contains the sample annotations:

  • SampleName: sample ID as encoded by the authors for the FragPipe search

  • PoolId: the TMT set identifier

  • Well: the well coordinate on the MS injection plate

  • Channel: the TMT reagent used for multiplexing

  • Chip: the N2 chip identifier

  • ChannelType: the type of channel, either single cells (n = 1988); bridge samples (n = 142); blank samples (n = 213) or blank samples that are isotopically adjacent to bridge samples(n = 142). Note that each set contains 2 bridge samples: one deuterated (TMT-135ND) and one non-deuterated (TMT-126).

  • LC: liquid chromatography column

  • RemovedByAuthors: FALSE if the sample ID is listed in the metadata file retrieved from the author's Github repository.

  • quantCols: used to format the input table as QFeatures object.

  • NPeptides: number of peptides identified in each sample

  • NProteins: number of proteins identified in each sample

  • MedianIntensity: median peptides intensity in each sample

Source

All data can be downloaded from the Zenodo repository. Data sources were identified from the preprint's data availability statement:

Data IdDescriptionSource
githubCode and metadatahttps://github.com/Cajun-data/RTLS-and-TMT32-scProteomics-of-PBMCs/tree/V1
msv000100684FragPipe outputftp://massive-ftp.ucsd.edu/v12/MSV000100684/
zenodo_18421469CellenOne outputhttps://zenodo.org/records/18421469
zenodo_18421672Authors intermediate datahttps://zenodo.org/records/18421672

Acquisition protocol

The data were acquired using the following setup. More information can be found in the source article (see References).

  • Sample collection: Blood samples taken from healthy donors and PBMCs were extracted.

  • Cell isolation: Cell were isolated using a cellenONE sorter and dispensed on in-house N2 nanoPOTS chips (array of 48 nanowells containing 32 inner wells each).

  • Sample preparation Single-cell protein lysates were digested overnight at 37°C in a digestion buffer containing DDM and 1/2 Lys-C/Trypsin. Subsequently, TMT reagents diluted in DMSO were added and incubated overnight. Labelling was quenched using HA with DTT. Sample were acidified using formic acid.

  • Liquid chromatography: Chromatographic separation of peptides was conducted using a home-built dual-column nanoPOTS autosampler. The LC consists of a parallel setup with an in-house packed solide phase extracton (100µm i.d., 4cm, 5µm C18) and an in-house pached LC column (50µm i.d., 25 cm, 1.7µm C18), operatd with an Ultimate 3000 RSLC nano Systems

  • Mass spectrometry: Acquisition was conducted with an Orbitrap Eclipse Tribrid with FAIMS. We refere to the original preprint for in-depth details on the MS settings

  • Raw data processing: Raw files were processed using FragPipe v23.0 (MSFragger v4.3; MSBooster v1.3.9; Percolator v3.7.1; Philosopher 5.1.1; IonQuant 1.11.11; TMT-Integrator 6.1.1). TMTpro reagents were set as static modificaton and M-oxidation as variable modification.

Data collection

The data were provided by the authors and is accessible at the following Zenodo repository. The repository contains the following files of interest:

The sample names were extracted from FragPipe's output table and some annotations were derived from the names (PoolId, Well, Channel, Chip, ChannelType). The LC information and which samples to remove was provided by the author as metadata and included in the annotations. The peptide table (abundance_peptide_None.tsv), along with the extracted annotations, were combined in a single QFeatures object using the scp::readSCP() function.

The protein table (abundance_protein_None.tsv) were formatted formated using QFeatures::readSummarizedExperiment() and added to the QFeatures object.

The peptide_modelled set was generated from the peptides set. First, the data were filtered to keep only single cells that were retained by the authors (cf batch effect discussed in Fig S4 in the preprint). Then, the data was minimally processed (encoding 0s as NA, log-transformation). Finally, an scplainer model was fit, which includes the median intensity as a normalisation factor, and TMT reagent and TMT set identifier as technical factor. An N/P threshold of 5 was applied.

References

Fulcher, J. M., Kwon, Y., Dawar, P., Kumar, R., Williams, S. M., Miller, P., ... Paša-Tolić, L. (2026). Single-cell proteomics of human peripheral blood mononuclear cells exceeding 600 cells per day. doi:10.64898/2026.02.01.703169

Examples

# \donttest{
fulcher2026()
#> see ?scpdata and browseVignettes('scpdata') for documentation
#> loading from cache
#> An instance of class QFeatures (type: scp) with 3 sets:
#> 
#>  [1] peptides: SummarizedExperiment with 8723 rows and 2485 columns 
#>  [2] proteins: SummarizedExperiment with 1661 rows and 2485 columns 
#>  [3] peptides_modelled: SingleCellExperiment with 8723 rows and 1055 columns 
# }