biostudies-arrayexpressMarc BeyerMus musculusbcl2fastq2 v2.20FastQC v0.11.9 , MultiQC v1.14, Snakemake v7.20.0, kallisto v0.48.0BD Chorushttps://www.ebi.ac.uk/biostudies/studies/E-MTAB-17301This experiment evaluates whether the gates predicted by the tool ConvexGating improve the isolation of adipocyte precursor subpopulations (P1 and P2) from one high-fat diet (HFD) mouse relative to conventional manual gating. Visceral white adipose tissue (WAT) was chosen as a challenging test case for algorithm transfer beyond immune cells, given its high lipid content and cellular fragility, which complicate accurate cell sorting. WAT cells were stained and sorted into P1 and P2 subpopulations, then subjected to plate-based single-cell RNA sequencing via FLASH-Seq to confirm subpopulation identity. By directly comparing the two gating strategies, we assessed whether ConvexGating yields superior population purity, robustness, and transcriptional resolution - and whether algorithm-guided gating more reliably captures functionally distinct adipocyte precursor states to enhance downstream transcriptomic profiling and biological interpretability.biostudies-arrayexpressNucleic Acid Extraction - White adipose tissue–derived adipocyte precursor subpopulations (P1 and P2) were isolated by fluorescence-activated cell sorting (FACS) into 384-well plates containing 1 µL of lysis buffer. The lysis buffer consisted of 0.2% Triton X-100, RNase inhibitor, dithiothreitol (DTT), betaine, and dNTPs. Cells were sorted directly into lysis buffer supplemented with a STRT-dT31 oligonucleotide and FLASH-UMI template switching oligo components. Plates were immediately sealed, snap-frozen on dry ice, and stored at −80°C until further processing.Sample Collection - Isolation and processing of visceral adipose tissue: The isolation of progenitor cells from the stromal vascular fraction (SVF) followed the protocol outlined by Sieckmann et al, Mol. Biol. Cell (2022). In brief, inguinal white adipose tissue (WAT) from high-fat diet-fed mice was excised, finely chopped, and enzymatically digested using 2 mg/mL collagenase II (Merck) and 15 kU/mL DNAse I (PanReac AppliChem) in PBS containing 0.5% bovine serum albumin (BSA; Sigma) at 37 °C under continuous agitation. The enzymatic reaction was terminated by adding AT buffer (PBS with 0.5% BSA). Following digestion, the cell suspension was filtered through a 100-μm strainer (Corning) and centrifuged at 500 × g for 10 min. The supernatant, which contained mature adipocytes, was removed, while the pellet comprising the stromal vascular fraction was resuspended in red blood cell lysis buffer (BioLegend) and incubated for 2 min at room temperature. This lysis step was halted by the addition of AT buffer, followed by centrifugation at 500 × g for 10 min. The isolated cells were subsequently kept in BD Omics Guard at 4°C overnight. On the next day, cells were washed, then stained with surface antibodies and incubated at 4°C for 30 minutes. Then, index sorting on a BD FACSDiscover S8 was performed.Library Construction - cDNA synthesis and amplification were performed following the FLASH-seq protocol with modifications for FLASH-UMI-seq. Reverse transcription and PCR amplification were carried out using Superscript IV reverse transcriptase with a template switching oligonucleotide containing unique molecular identifiers (UMIs) and a 5 bp spacer. RT-PCR conditions consisted of reverse transcription at 50°C for 60 minutes, initial denaturation at 98°C for 3 minutes, followed by 21 cycles of amplification (98°C for 20 s, 65°C for 20 s, and 72°C for 6 min). Amplified cDNA was purified using magnetic bead-based cleanup. Libraries were generated from normalized cDNA (~200 pg/µL) using Tn5 transposase-based tagmentation with in-house loaded Tn5 enzyme. Library enrichment PCR was performed using indexed primers compatible with Illumina sequencing. Final libraries were quantified using Qubit dsDNA HS assay and fragment size distribution was assessed using Agilent TapeStation D1000.Sequencing - Pooled libraries were sequenced in paired-end mode (2 × 101 bp) on an Illumina NovaSeq 6000 system using S1 v1.5 chemistry. Base calling and demultiplexing were performed using Illumina bcl2fastq v2.20 to generate FASTQ files. Index reads (i7/i5) were used for sample demultiplexing. Read 1 contained the UMI and 5′ end of the transcript, while Read 2 contained the corresponding cDNA fragment sequence.OrganizationMINSEQE ScoreAssays and DataProcessed DataMAGE-TAB FilesData Transformation - Single-cell RNA-seq data from mouse white adipose tissue–derived adipocyte precursor subpopulations (P1 and P2) were demultiplexed using bcl2fastq2 (v2.20). Raw sequencing reads were subjected to initial quality control using FastQC (v0.11.9) and MultiQC (v1.14). Processed reads were pseudoaligned to the mouse transcriptome (GRCm38/Gencode M25 or equivalent reference used in analysis) using Kallisto (v0.48.0) with default parameters, including an estimated average fragment length of 200 bp and a standard deviation of 30 bp. The alignment workflow was implemented using Snakemake (v7.20.0). Transcript abundance estimates generated by Kallisto in HDF5 format were merged and converted into gene-level expression matrices using tximport (v1.20.0). Expression values were normalized to Transcripts Per Million (TPM) and further transformed using length-scaled TPM (“lengthScaledTPM”) with default settings. Gene identifiers were converted from Ensembl IDs to gene symbols using biomaRt (v2.48.3). The resulting expression matrix was imported into a Seurat object for downstream analysis.MetabolomicsUnknownTranscriptomicsGenomicsProteomicsI.DOT (Dispendix), C.WASH (CYTENA), Qubit Fluorometer (Thermo Fisher Scientific), Agilent TapeStation 4200Illumina NovaSeq 6000BD FACSDiscover S8RNA-seq of coding RNA from single cellsMus musculusConvexGating infers gating strategies from clusters in single cell cytometry dataMarc BeyerVincent D. Friedrich, Karola Mai, Thomas P. Hofer, Elfriede Nößner, Lorenzo Bonaguro, Celia L. Hartmann, Aleksej Frolov, Caterina Carraro, Doaa Hamada, Mehrnoush Hadaddzadeh-Shakiba, Heidi Theis, Dalila Juliana Silva Ribeiro, Dagmar Wachten, F. Thomas Wunderlich, Markus Scholz, Fabian J. Theis, Matthias Becker, Marc D. Beyer Joachim L. Schultze, Maren BüttnerfalseFLASH-seq analysis of adipocyte precursor subpopulations (P1 and P2) isolated from a high-fat diet (HFD) mouse, comparing two sorting strategies: conventional manual gating and ConvexGatingThis experiment evaluates whether the gates predicted by the tool ConvexGating improve the isolation of adipocyte precursor subpopulations (P1 and P2) from one high-fat diet (HFD) mouse relative to conventional manual gating. Visceral white adipose tissue (WAT) was chosen as a challenging test case for algorithm transfer beyond immune cells, given its high lipid content and cellular fragility, which complicate accurate cell sorting. WAT cells were stained and sorted into P1 and P2 subpopulations, then subjected to plate-based single-cell RNA sequencing via FLASH-Seq to confirm subpopulation identity. By directly comparing the two gating strategies, we assessed whether ConvexGating yields superior population purity, robustness, and transcriptional resolution - and whether algorithm-guided gating more reliably captures functionally distinct adipocyte precursor states to enhance downstream transcriptomic profiling and biological interpretability.2026-07-17T00:00:00Z2026-07-17T01:00:40.008Z2026-07-16T11:29:55.357ZE-MTAB-17301ERP201313EFO_0002944EFO_0004170EFO_0005684EFO_0005518EFO_0003816EFO_0004184