Anne Bertolini, Michael Prummer, Mustafa Anil Tuncel, Ulrike Menzel, María Lourdes Rosano-González, Jack Kuipers, Daniel Johannes Stekhoven, Tumor Profiler consortium , Niko Beerenwinkel, Franziska Singer
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.
In recent years, single-cell RNA sequencing (scRNA-seq) emerged as a high-throughput technology for uncovering gene expression at the single-cell level, which provides unprecedented insights into, e.g., cell differentiation, the immune compartment, and tumor heterogeneity [1,2]. Initially used to characterize PBMCs or differentiating stem cells, an increasing number of studies exploit scRNA-seq to investigate clinical samples such as tumor tissues [3,4]. There are multiple software suites available with extensive functionality for general scRNA-seq analysis, including the widely-used tools SEURAT  and ScanPy  or the web-based software suites CreSCENT  and ASAP . However, they have some disadvantages: First, for non-bioinformaticians the usage can be difficult because setting all parameters and applying the different steps requires at least basic R or Python programming knowledge. Second, to the best of our knowledge, no software is available that facilitates in-silico drug candidate identifications based on single-cell data. Finally, existing software suites are not designed to manage large-scale data analysis in a highly reproducible, transparent, and auditable way, including error tracking and process documentation, and thus are not suitable to be employed for routine clinical use [9,10].
We showcase the readout and analyses possible with scAmpi for scRNA-seq data from a biopsy of a melanoma patient who was included in the Tumor Profiler clinical study . The full analysis from raw fastq files to in-silico drug candidate identification is triggered with only two commands. For details on the default parameter settings, we refer to S1 Text. In the initial mapping step, Cellranger identifies 4193 cells. Subsequent filtering in scAmpi removes 10% (437) of cells due to low quality (Fig 2A). Fig 2B and 2C show examples of QC metrics on the UMAP representation of the cells. After normalization, the cell-cycle phase has no apparent effect on the embedding of the cells anymore. Instead, as shown in Fig 3A, the embedding is cleanly separated by cell type populations.
Membership of the Tumor Profiler consortium:
Rudolf Aebersold, MelikeAk, Faisal S Al-Quaddoomi, Jonas Albinus, IlariaAlborelli, SonaliAndani, Per-OlofAttinger, Marina Bacac, Daniel Baumhoer, Beatrice Beck-Schimmer, Niko Beerenwinkel, Christian Beisel, Lara Bernasconi, Anne Bertolini, Bernd Bodenmiller, Ximena Bonilla, Lars Bosshard, Byron Calgua, Ruben Casanova, StéphaneChevrier, Natalia Chicherova, Maya D’Costa, Esther Danenberg, Natalie Davidson, Monica-AndreeaDrăgan, ReinhardDummer, Stefanie Engler, Martin Erkens, KatjaEschbach, Cinzia Esposito, André Fedier, Pedro Ferreira, Joanna Ficek, Anja L Frei, Bruno Frey, Sandra Goetze, Linda Grob, Gabriele Gut, DetlefGünther, Martina Haberecker, PirminHaeuptle, Viola Heinzelmann-Schwarz, Sylvia Herter, Rene Holtackers, Tamara Huesser, AnjaIrmisch, Francis Jacob, Andrea Jacobs, Tim M Jaeger, Katharina Jahn, Alva R James, Philip M Jermann, André Kahles, Abdullah Kahraman, Viktor H Koelzer, Werner Kuebler, Jack Kuipers, Christian P Kunze, Christian Kurzeder, Kjong-Van Lehmann, Mitchell Levesque, Sebastian Lugert, GerdMaass, Markus G Manz, Philipp Markolin, Julien Mena, Ulrike Menzel, Julian M Metzler, Nicola Miglino, Emanuela S Milani, HolgerMoch, Simone Muenst, Riccardo Murri, Charlotte KY Ng, Stefan Nicolet, Marta Nowak, Patrick GA Pedrioli, Lucas Pelkmans, Salvatore Piscuoglio, Michael Prummer, Mathilde Ritter, Christian Rommel, María L Rosano-González, Gunnar Rätsch, NataschaSantacroce, JacoboSarabia del Castillo, Ramona Schlenker, Petra C Schwalie, Severin Schwan, Tobias Schär, Gabriela Senti, Franziska Singer, SujanaSivapatham, BerendSnijder, Bettina Sobottka, Vipin T Sreedharan, Stefan Stark, Daniel J Stekhoven, Alexandre PA Theocharides, Tinu M Thomas, Markus Tolnay, VinkoTosevski, Nora C Toussaint, Mustafa A Tuncel, Marina Tusup, Audrey Van Drogen, Marcus Vetter, TatjanaVlajnic, Sandra Weber, Walter P Weber, RebekkaWegmann, Michael Weller, Fabian Wendt, Norbert Wey, Andreas Wicki, Mattheus HE Wildschut, Bernd Wollscheid, Shuqing Yu, Johanna Ziegler, Marc Zimmermann, Martin Zoche, Gregor Zuend.
Citation: Bertolini A, Prummer M, Tuncel MA, Menzel U, Rosano-González ML, Kuipers J, et al. (2022) scAmpi—A versatile pipeline for single-cell RNA-seq analysis from basics to clinics. PLoSComputBiol 18(6): e1010097. https://doi.org/10.1371/journal.pcbi.1010097
Editor: Jason A. Papin, University of Virginia, UNITED STATES
Received: April 26, 2021; Accepted: April 12, 2022; Published: June 3, 2022
Copyright: © 2022 Bertolini et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The source code of scAmpi as well as usage examples are distributed open source on github at https://github.com/ETH-NEXUS/scAmpi_single_cell_RNA.
Funding: The study described in this paper is the result of a jointly-funded effort between several academic institutions (University of Zurich, University Hospital Zurich, Swiss Federal Institute of Technology in Zurich, University Hospital Basel), as well as F. Hoffmann-La Roche AG. They were involved in data collection. UM was supported by the ETH domain Personalized Health and Related Technologies (PHRT-510). The PHRT had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.