One-Step Cleanup for Single-Cell RNA Sequencing of Human Tumor Tissue
App Note / Case Study
Published: September 24, 2026
Brought to you by
Miltenyi Biotec
For more than 30 years, Miltenyi Biotec has played an important role in the design, development, manufacture and integration of products that empower the advancement of biomedical research and enable cell and gene therapy. They believe in biomedical science as the strongest driver to change the way we treat cancer, autoimmune diseases and neurodegenerative disorders.
Credit: Miltenyi Biotec
Single-cell RNA sequencing helps profile cell types in a tumor. But dead cells, red blood cells, and cell-free organelles in a dissociated sample add unwanted reads that can complicate those assignments.
A magnetic bead protocol now reduces all three contaminants in one cleanup step. In a direct comparison, it left more cells viable than conventional two-step cleanup with red blood cell lysis.
This app note measures cell and sequencing quality in Fresh human ovarian tumor samples. It then tests cryopreserved samples, where the magnetic protocol saved 25 min.
Download this app note to learn:
- Why higher viability and staining efficiency matter for sequencing readiness
- How fewer hemoglobin and mitochondrial reads support the study of rare cell populations
- Which cleanup method kept gene expression closest to the unprocessed sample
Cleaner samples, clearer data
A one-step sample cleanup workflow
for high-quality single-cell RNA sequencing
Sample preparation is fundamental to the success of any
single-cell RNA sequencing (scRNA-seq) experiment.
It determines the quality and reliability of downstream data and
thus the success of the experiment. In addition, well-prepared
samples can have a major impact on the reproducibility,
reliability, and actual cost of the whole experiment, while high
technical noise due to poor sample preparation can affect
data interpretation and, in the worst case, lead to unwanted
misinterpretations and false conclusions. The solution to
prevent this is to effectively remove dead cells, red blood cells
(RBCs), and debris. Here, we describe a one-step workflow that
combines several important cleaning steps into a single step,
significantly speeding up the sample cleaning process while
maintaining high sample quality without negatively impacting
the transcriptome.
This application note demonstrates the effectiveness of the
Cell Suspension Cleanup Kit, human (CSCK)*, on fresh and
cryopreserved human high-grade serous ovarian carcinoma
(HGSOC) samples that were dissociated with the gentleMACS™
Technology using the Tumor Dissociation Kit, human.
The quality of the scRNA-seq data was thus improved, as
demonstrated by several quality control (QC) parameters.
These include median gene or UMI counts per cell and reduced
cell-free content, such as mitochondrial reads. We also compared
this one-step CSCK workflow with a commercially available
two-step alternative. Using whole exome RNA sequencing
data, we identified significant transcriptional changes. In short,
the one-step method not only resulted in higher viability and
improved input material for high-quality scRNA-seq data, but
also reduced susceptibility to transcriptional alterations during
the cleanup process.
Materials and methods
Sample origin information
HGSOC samples were kindly provided by Evangelisches
Krankenhaus Frauenklinik Bergisch Gladbach, Germany
(Prof. Dr. Rudlowski). All patients included in this study gave
informed consent (ethical vote medical association Nordrhein:
2024026). Whole blood samples were withdrawn from healthy
donors who gave informed consent (ethical vote medical
association Nordrhein: 2020272).
Long-term storage of tumor samples
HGSOC samples were cryopreserved as tissue pieces in
MACS® Freezing Solution and stored in liquid nitrogen for
6 to 17 months for later bulk RNA-seq experiment.
Tumor dissociation
For scRNA-seq: Freshly obtained HGSOC samples were
dissociated immediately after acquisition using the
gentleMACS Octo Dissociator with Heaters and the Tumor
Dissociation Kit, human, in combination with gentleMACS
C Tubes. After tissue dissociation, single-cell suspensions were
filtered through MACS SmartStrainers (70 µm) and further
processed for cleanup and scRNA-seq.
For bulk RNA sequencing: Cryopreserved tumor samples were
thawed and dissociated using the gentleMACS Octo Dissociator
with Heaters and the Tumor Dissociation Kit, human,
in combination with gentleMACS C Tubes. After tissue
dissociation, single-cell suspensions were filtered through
MACS SmartStrainers (70 µm). To measure standard QC
parameters prior to sequencing experiments, an aliquot of cells
was used for flow cytometry on the MACSQuant® 10 Analyzer
and Countess™ 3 FL Automated Cell Counter (Thermo Fisher
Scientific®).
* Please note, the Cell Suspension Cleanup Kit, human, is not compatible with digestion enzymes degrading glycopherin A, including the following
Miltenyi Biotec Kits: Brain Tumor Dissociation Kit (P), human (130-095-942); Umbilical Cord Dissociation Kit, human (130-105-737); Embryoid Body Dissociation
Kit, human and mouse (130-096-348); Whole Skin Dissociation Kit, human (130-101-540); Multi Tissue Dissociation Kit 2 (130-110-203). As a workaround for these
samples, you can deplete RBCs using the Red Blood Cell Lysis Solution and subsequently use either the Cell Suspension Cleanup Kit, human, to deplete dead
cell and organelles, or, if only dead cells shall be removed, the Dead Cell Removal Kit.
1
Sample cleanup
scRNA-seq methods
For scRNA-seq: After dissociation, single-cell suspensions were
divided into two fractions. One fraction was the non-processed
origin fraction (Ori) and the second one was purified using the
workflow with CSCK.
For bulk RNA sequencing: After dissociation, whole blood
was added to the single-cell suspension for all samples to
compensate for the loss of erythrocytes during the freeze-thaw
cycle, after which the sample was divided into three fractions.
RNA of one unprocessed fraction was immediately isolated
and was later compared to the second and third fraction.
The second fraction was processed using the one-step cleanup
workflow with CSCK. The third fraction was cleaned up using
a two-step workflow with commercially available products,
including a column-free dead cell removal solution and a red
blood cell lysis solution. All samples were processed according
to the manufacturer’s instructions.
Flow cytometry analysis
After treatment, all samples were stained for a viability check.
Three samples were stained for testing erythrocyte and
mitochondrial content using anti-human antibody conjugates
manufactured by Miltenyi Biotec.
The following reagents were used:
CD45-VioGreen™, Annexin V-FITC Kit, CD326 (EpCAM)-PE,
CD31-PE-Vio® 770, CD90-APC, CD235a(GlyA)-APC Vio 770,
DAPI, and propidium iodide (PI).
Flow cytometry analysis was performed using the
MACSQuant Analyzer 10.
Measurement of staining efficiency
In parallel to flow analysis, a sample aliquot was used to
validate sample count, viability, and staining efficiency on
the Countess 3 FL Automated Cell Counter using ReadyCount™
Green/Red Viability Stain following manufacturer’s instructions
(Thermo Fisher Scientific). For optimal NGS results, the research
community commonly recommends a staining efficiency of
70–80%. Furthermore, 10× Genomics® recommends a maximum
RBC content of 20% in the sample.1
Sample preparation
Fresh tissue
Dissociation
Filter
Single-cell suspensions processed with CSCK and unprocessed
samples as control from three freshly obtained HGSOC samples
were resuspended in PBS with 0.04% non-acetylated BSA.
They were then subsequently processed using the Chromium™
Next GEM Single Cell 5' Reagent Kit (Dual Index) according to
the manufacturer’s instructions (10× Genomics).
CSCK-processed and unprocessed samples from each tumor
donor were sequenced to a depth of >20,000 reads per cell
using a NextSeq® 2000 Sequencing System (Illumina®).
Sequencing data were processed using the Cell Ranger
software and the results were visualized using the Loupe
Cell Browser software (both 10x Genomics).
Whole transcriptome analysis
RNA was isolated from three tumor samples using the RNeasy®
Kit (QIAGEN®) and quantified using Qubit™ 4 Fluorometer
(Thermo Fisher Scientific). Sample quality was determined with
an automated electrophoresis tool using the 2100 Bioanalyzer®
Instrument (Agilent Technologies). The following library
preparation was performed using QIAseq™ Stranded mRNA
Library Kit and HiFi PCR Master Mix (QIAGEN). The library was
sequenced on the NextSeq 2000 sequencer. Transcriptome
sequencing data analysis was performed via CLC Genomics
Workbench/Server version 22.0 (QIAGEN) covering trimming of
reads, read mapping (hg38, forward strand only), differential
gene expression analysis (‘While controlling for = Donor’), and
principal component analysis. Genes with a false discovery rate
of 1.5 were considered to be differentially expressed.
Results
One-step quality improvement
Single-cell suspensions were prepared from five freshly
obtained HGSOC samples following the process described in
the materials and methods section and also shown in figure 1.
One-step cleanup
Label
Wash
Downstream analysis
Sequencing
A C G T
Figure 1: Schematic representation of the sample processing workflow. Samples can be stored or transported in MACS Tissue Storage Solution
before undergoing tissue dissociation with the gentleMACS Octo Dissociator with Heaters. The resulting single-cell suspension is subjected to a
one-step magnetic cleanup with CSCK to remove debris, dead cells, red blood cells (RBCs), and cell-free organelles. This ensures high-quality input for
downstream applications such as bulk or scRNA-seq.
2
Next, the reads that were confidently mapped to intergenic
and transcriptomic regions were analyzed (figs. 3B and C).
While fewer intergenic regions were detected, overall a much
higher proportion of reads were confidently mapped to the
transcriptome in the CSCK-processed sample. This indicates
more relevant data generated in the scRNA-seq experiment
after CSCK purification compared to the control (Ori).
transcriptome (n = 3).
Figure 3: Optimizing sequencing results of freshly obtained
HGSOC samples. scRNA-seq quality metrics comparing unprocessed
(Ori) and samples cleaned with CSCK. (A) Distribution of valid barcodes.
(B) Distribution of reads mapped confidently to intergenic (non-coding)
regions, and (C) proportion of reads mapped confidently to the
Ori CSCK
Reads
transcriptome (%)
50
55
60
65
70
C
75
Valid barcodes (%)
87
88
89
90
91
92
scRNA-seq of single-cell suspensions from three of the five
freshly obtained HGSOC samples was performed on
unprocessed and purified cells using CSCK. Various NGS quality
parameters were plotted to highlight the effect on data
reliability when a sample is properly prepared. The proportion
of valid barcodes differed between the samples cleaned with
CSCK and the unprocessed samples, with higher values in the
cleaned sample resulting in a better match to the barcode
whitelist. The unprocessed sample had more sequencing
errors, reducing the proportion of valid barcodes.
A
Cleanedup samples improve scRNA-seq results
Figure 2: Cleanup performance of freshly obtained HGSOC
samples: (A) Viability before and after processing the sample with
CSCK. (B) Frequency shows the frequency of RBCs with a threshold
set at 20%, as recommended for single-cell RNA sequencing1, and
(C) the frequency of cell-free organelles (mitochondria). (D) Staining
efficiency, which is a critical factor for subsequent successful single-
cell RNA sequencing. Dotted lines represent the respective threshold
of what has been commonly used as a quality indication for viability,
RBC content, and staining efficacy (n = 5).
3
0
Reads intergenic
regions (%)
2
4
6
B
8
With CSCK, all relevant quality parameters were increased after
the one-step magnetic separation (cleanup) run compared to
the untreated sample. Viability was increased from 52% to 77%
(fig. 2A), while RBC and cell-free organelle content was reduced
to 6.8% and 7.4%, respectively (fig. 2B and 2C). This ultimately
resulted in an increase in overall staining efficiency measured
with the Countess 3 FL from 37% to 89% (fig. 2D), representing
the percentage of viable nucleated cells. Importantly, sample
cleanup with CSCK resulted in compliance with the thresholds,
indicating that the sample was well prepared. Moreover, the
staining efficiency value of the automated cell counter is a
decisive factor in determining the suitability for the subsequent
sequencing. The higher the value, the more predictive it is of
sequencing success. Ori CSCK
Frequency (%)
20
0
Mitochondria
40
60
80
Staining efficiency (%)
20
0
40
60
80
C
100
100
D
0
Viable cells
0
Quality parameters were assessed using the MACSQuant 10
Analyzer and Countess 3 FL in conjunction with the
ReadyCount Green/Red Viability Stain. The proportion of
dead cells was analyzed using the viability marker PI and RBCs
using GlyA antibodies. Tom22 was selected as a marker for
mitochondria, which were determined to be representative
of cell-free organelles (fig. 2). The enriched sample (CSCK) was
compared to the original unpurified fraction (Ori) immediately
after dissociation.
Frequency (%)
20
40
60
80
Frequency (%)
20
40
60
80
A
100
100
B
RBCs
4
Removal of RBCs and cell-free organelles
increases readout quality
In addition to dead cells, the CSCK also effectively removes
RBCs and cell-free organelles, like mitochondria and cell-free
nuclei, which can have a huge impact on the quality of the
sequencing data. To demonstrate this, scRNA-seq was
performed with unprocessed (Ori) and CSCK-processed freshly
obtained HGSOC samples. Figure 4A presents a representative
result from a CSCK-processed sample showing a UMAP
projection with general cell type annotation. Resulting cell
populations exhibit clear separation, enabling detailed
downstream analysis, including the identification and
investigation of rare cell subsets. Additionally, RBC-associated
gene expression (HBB+) was analyzed, and the number of
positive reads was compared between the Ori and CSCK (fig. 4B).
A significantly higher number of HBB+ reads were detected
in the unprocessed (Ori) sample, potentially confounding cell
population assignments and leading to misinterpretations.
CSCK-processed samples, show a strong reduction of these
reads. Another crucial factor of technical noise can be caused
by a high mitochondrial (mt) read count. Figure 4C shows that
CSCK was able to significantly reduce this amount compared
to the unprocessed sample, confirming the result shown in
figure 2C.
Reducing unwanted sequencing reads enables a lower read
threshold, allowing for a more detailed and comprehensive
view of the data, such as the analysis of rare cell populations.
Figure 4: Enhancing output quality and reducing unwanted
sequencing reads of freshly obtained HGSOC. (A) UMAP plot
of a CSCK-processed sample with annotated cell types based on
transcriptomic data. (B) UMAP plots comparing single-cell RNA
sequencing data from unprocessed (Ori) and CSCK-processed
samples (CSCK), highlighting hemoglobin (HBB)-positive reads in
red that interfere with other populations as seen in the Ori sample.
(C) Percentage of mitochondrial (mt) reads in unprocessed (Ori) and
CSCK-processed samples, demonstrating a significant reduction in
mt reads in CSCK-processed samples (n = 3).
UMAP 2
UMAP 1
10
0-5-10-10 0 10
5
15
A
Adipocytes
B cells
CD4+ T cells
CD8+ T cells
Endothelial cells
Monocytes
NK cells
Other
Epithelial cells
Erythrocytes
Fibroblasts
Macrophages
Mesangial cells
B
UMAP 2
UMAP 1
10
0-5-10-10 0 10
5
15
Ori CSCK-10 0 10
Download the Application Note for FREE Now!
Information you provide will be shared with the sponsors for this content. Technology Networks or its sponsors may contact you to offer you content or products based on your interest in this topic. You may opt-out at any time.
Experiencing issues viewing the form? Click here to access an alternate version