scAmpi - A versatile pipeline for single-cell RNA-seq analysis from basics to clinics

Anne Bertolini, Michael Prummer, Anıl Tuncel, Ulrike Menzel, María L. Rosano-González, Jack Kuipers, Daniel J. Stekhoven, Niko Beerenwinkel, Franziska SingerbioRxiv 2021 · DOI

Abstract

Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful technique to decipher tissue composition at the single-cell level and to inform on disease mechanisms, tumor heterogeneity, and the state of the immune microenvironment. Although multiple methods for the computational analysis of scRNA-seq data exist, their application in a clinical setting demands standardized and reproducible workflows, targeted to extract, condense, and display the clinically relevant information. To this end, we designed scAmpi (Single Cell Analysis mRNA pipeline), a workflow that facilitates scRNA-seq analysis from raw read processing to informing on sample composition, clinically relevant gene and pathway alterations, and in silico identification of personalized candidate drug treatments. We demonstrate the value of this workflow for clinical decision making in a molecular tumor board as part of a clinical study.

Introduction: The Clinical Gap in Single-Cell Analysis

Single-cell RNA sequencing has revolutionized our ability to see the "fine print" of tumors, allowing researchers to identify specific cell types and immune interactions that bulk sequencing misses. However, moving this technology into routine clinical use faces three major hurdles:

  1. Complexity: Popular tools like Seurat or Scanpy require significant R or Python programming knowledge, making them inaccessible to many clinicians.
  2. Lack of Standardization: Clinical settings require "auditable" workflows where every step is documented and reproducible. Most current research scripts are too "ad-hoc."
  3. Actionability: While existing tools can tell you what cells are in a tumor, they rarely tell you what to do about it. There is a gap between identifying a gene expression pattern and identifying a potential drug.

The scAmpi Workflow

The authors developed scAmpi (Single Cell Analysis mRNA pipeline) using the Snakemake workflow management system. This choice ensures that the pipeline is modular, scalable, and generates detailed log files for every step—essential for clinical transparency and reproducibility.

The scAmpi end-to-end workflow, from raw transcripts to drug prediction.

The pipeline follows a logical progression from raw data to clinical insight:

  • Preprocessing: It uses Cellranger for initial mapping and sctransform for normalization. A critical step involves filtering out low-quality cells, such as those with high mitochondrial DNA content, which often indicates cell death.
  • Cell Typing: It employs a hierarchical, marker-based classification system. It first identifies major populations (e.g., Tumor vs. T cell) and then drills down into specific subtypes (e.g., Regulatory T cells).
  • Pathway Analysis: To avoid the bias of needing a "reference" population, scAmpi uses GSVA (Gene Set Variation Analysis), which ranks pathways within each individual cell.
  • Drug Prediction: This is the most novel component. The pipeline queries the DGIdb (Drug-Gene Interaction Database) and enriches the results with curated data from CIViC and clinicaltrials.gov. This allows the system to suggest drugs based on whether a specific gene is over- or under-expressed in the patient's tumor clusters.

Quality Control and Preprocessing

A cornerstone of clinical-grade analysis is rigorous quality control. scAmpi automates the detection of technical artifacts and biological noise. As shown in the QC metrics below, the pipeline identifies cells with high mitochondrial (MT) read fractions and low gene detection counts, which are typically excluded to ensure the downstream analysis is based on viable, high-quality cells. Furthermore, the pipeline accounts for the cell cycle phase, ensuring that proliferative states do not confound the identification of distinct cell types.

Quality control metrics in scAmpi, including mitochondrial read fraction filtering and cell cycle phase assignment.

Key Results and Clinical Validation

The paper validates scAmpi using a real-world case study of a melanoma biopsy from the "Tumor Profiler" clinical study. Results were compared against other high-tech methods like CyTOF (mass cytometry) and Pharmacoscopy (ex vivo drug response testing) to ensure accuracy.

  • Efficiency: A standard sample of 4,000 cells can be processed from start to finish in 4 to 8 hours, making it fast enough for clinical turnaround times.
  • Biological Accuracy: In the melanoma case study, scAmpi correctly identified that the tumor's MAPK pathway was down-regulated, correctly predicting that standard BRAF/MEK inhibitors would likely fail.
  • Actionable Discovery: The pipeline identified an over-expression of CCND1 and CDK4 in the tumor cells. It flagged palbociclib as a potential treatment—a finding that was independently confirmed by the study’s functional drug-response testing (Pharmacoscopy).

Discussion and Future Directions

scAmpi represents a major step toward bridging the "drug gap" in single-cell genomics. By using Snakemake, the authors ensure that an analysis performed today can be exactly replicated months later, meeting a core requirement for clinical certification. While traditional pathway analysis often requires a control group, scAmpi’s use of GSVA allows for the analysis of a single sample in isolation by comparing gene sets against each other within the same cell.

While the current default settings are optimized for 10x Genomics data, the modular nature of the pipeline allows for other mappers to be swapped in. The authors have open-sourced the code on GitHub, inviting the community to contribute more tissue-specific marker lists and drug-interaction modules. Full integration into hospital IT systems and regulatory approval (like IVDR in Europe) remains the next major challenge for the field.

Summary updated

Abstract

Abstract Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful technique to decipher tissue composition at the single-cell level and to inform on disease mechanisms, tumor heterogeneity, and the state of the immune microenvironment. Although multiple methods for the computational analysis of scRNA-seq data exist, their application in a clinical setting demands standardized and reproducible workflows, targeted to extract, condense, and display the clinically relevant information. To this end, we designed scAmpi ( S ingle C ell A nalysis m RNA pi peline), a workflow that facilitates scRNA-seq analysis from raw read processing to informing on sample composition, clinically relevant gene and pathway alterations, and in silico identification of personalized candidate drug treatments. We demonstrate the value of this workflow for clinical decision making in a molecular tumor board as part of a clinical study.
BibTeX
@misc{bertolini2021scampi,
  author = {Anne Bertolini and Michael Prummer and Anıl Tuncel and Ulrike Menzel and María L. Rosano-González and Jack Kuipers and Daniel J. Stekhoven and Niko Beerenwinkel and Franziska Singer},
  title = {{scAmpi - A versatile pipeline for single-cell RNA-seq analysis from basics to clinics}},
  year = {2021},
  doi = {10.1101/2021.03.25.437054},
  howpublished = {bioRxiv},
}
Figure