hadge: a comprehensive pipeline for donor deconvolution in single-cell studies.

Single-cell multiplexing techniques (cell hashing and genetic multiplexing) combine multiple samples, optimizing sample processing and reducing costs. Cell hashing conjugates antibody-tags or chemical-oligonucleotides to cell membranes, while genetic multiplexing allows to mix genetically diverse samples and relies on aggregation of RNA reads at known genomic coordinates. We develop hadge (hashing deconvolution combined with genotype information), a Nextflow pipeline that combines 12 methods to perform both hashing- and genotype-based deconvolution. We propose a joint deconvolution strategy combining best-performing methods and demonstrate how this approach leads to the recovery of previously discarded cells in a nuclei hashing of fresh-frozen brain tissue.

Ex vivo tissue perturbations coupled to single-cell RNA-seq reveal multilineage cell circuit dynamics in human lung fibrogenesis.

Pulmonary fibrosis develops as a consequence of failed regeneration after injury. Analyzing mechanisms of regeneration and fibrogenesis directly in human tissue has been hampered by the lack of organotypic models and analytical techniques. In this work, we coupled ex vivo cytokine and drug perturbations of human precision-cut lung slices (hPCLS) with single-cell RNA sequencing and induced a multilineage circuit of fibrogenic cell states in hPCLS. We showed that these cell states were highly similar to the in vivo cell circuit in a multicohort lung cell atlas from patients with pulmonary fibrosis. Using micro-CT-staged patient tissues, we characterized the appearance and interaction of myofibroblasts, an ectopic endothelial cell state, and basaloid epithelial cells in the thickened alveolar septum of early-stage lung fibrosis. Induction of these states in the hPCLS model provided evidence that the basaloid cell state was derived from alveolar type 2 cells, whereas the ectopic endothelial cell state emerged from capillary cell plasticity. Cell-cell communication routes in patients were largely conserved in hPCLS, and antifibrotic drug treatments showed highly cell type-specific effects. Our work provides an experimental framework for perturbational single-cell genomics directly in human lung tissue that enables analysis of tissue homeostasis, regeneration, and pathology. We further demonstrate that hPCLS offer an avenue for scalable, high-resolution drug testing to accelerate antifibrotic drug development and translation.

An integrated cell atlas of the lung in health and disease.

Single-cell technologies have transformed our understanding of human tissues. Yet, studies typically capture only a limited number of donors and disagree on cell type definitions. Integrating many single-cell datasets can address these limitations of individual studies and capture the variability present in the population. Here we present the integrated Human Lung Cell Atlas (HLCA), combining 49 datasets of the human respiratory system into a single atlas spanning over 2.4 million cells from 486 individuals. The HLCA presents a consensus cell type re-annotation with matching marker genes, including annotations of rare and previously undescribed cell types. Leveraging the number and diversity of individuals in the HLCA, we identify gene modules that are associated with demographic covariates such as age, sex and body mass index, as well as gene modules changing expression along the proximal-to-distal axis of the bronchial tree. Mapping new data to the HLCA enables rapid data annotation and interpretation. Using the HLCA as a reference for the study of disease, we identify shared cell states across multiple lung diseases, including SPP1+ profibrotic monocyte-derived macrophages in COVID-19, pulmonary fibrosis and lung carcinoma. Overall, the HLCA serves as an example for the development and use of large-scale, cross-dataset organ atlases within the Human Cell Atlas.

mlf-core: a framework for deterministic machine learning.

MOTIVATION: Machine learning has shown extensive growth in recent years and is now routinely applied to sensitive areas. To allow appropriate verification of predictive models before deployment, models must be deterministic. Solely fixing all random seeds is not sufficient for deterministic machine learning, as major machine learning libraries default to the usage of nondeterministic algorithms based on atomic operations. RESULTS: Various machine learning libraries released deterministic counterparts to the nondeterministic algorithms. We evaluated the effect of these algorithms on determinism and runtime. Based on these results, we formulated a set of requirements for deterministic machine learning and developed a new software solution, the mlf-core ecosystem, which aids machine learning projects to meet and keep these requirements. We applied mlf-core to develop deterministic models in various biomedical fields including a single-cell autoencoder with TensorFlow, a PyTorch-based U-Net model for liver-tumor segmentation in computed tomography scans, and a liver cancer classifier based on gene expression profiles with XGBoost. AVAILABILITY AND IMPLEMENTATION: The complete data together with the implementations of the mlf-core ecosystem and use case models are available at https://github.com/mlf-core.

Best practices for single-cell analysis across modalities.

Recent advances in single-cell technologies have enabled high-throughput molecular profiling of cells across modalities and locations. Single-cell transcriptomics data can now be complemented by chromatin accessibility, surface protein expression, adaptive immune receptor repertoire profiling and spatial information. The increasing availability of single-cell data across modalities has motivated the development of novel computational methods to help analysts derive biological insights. As the field grows, it becomes increasingly difficult to navigate the vast landscape of tools and analysis steps. Here, we summarize independent benchmarking studies of unimodal and multimodal single-cell analysis across modalities to suggest comprehensive best-practice workflows for the most common analysis steps. Where independent benchmarks are not available, we review and contrast popular methods. Our article serves as an entry point for novices in the field of single-cell (multi-)omic analysis and guides advanced users to the most recent best practices.

A Mutation-Agnostic Hematopoietic Stem Cell Gene Therapy for Metachromatic Leukodystrophy.

Metachromatic leukodystrophy (MLD) is a rare genetic disorder caused by mutations in the Arylsulfatase-A (ARSA) gene. The enzyme plays a key role in sulfatide metabolism in brain cells, and its deficiency leads to neurodegeneration. The clinical manifestations of MLD include stagnation and decline of motor and cognitive function, leading to premature death with limited standard treatment options. Here, we describe a mutation-agnostic hematopoietic stem and progenitor cell (HSPC) gene therapy using CRISPR-Cas9 and AAV6 repair template as a prospective treatment option for MLD. Our strategy achieved efficient insertions and deletions (>87%) and a high level of gene integration (>47%) at the ARSA locus in human bone marrow-derived HSPCs, with no detectable off-target editing. As a proof of concept, we tested our mutation-agnostic therapy in HSPCs derived from two MLD patients with distinct mutations and demonstrated restoration of ARSA enzyme activity (>30-fold improvement) equivalent to healthy adults. In summary, our investigation enabled a mutation-agnostic therapy for MLD patients with proven efficacy and strong potential for clinical translation.

Sfaira accelerates data and model reuse in single cell genomics.

Single-cell RNA-seq datasets are often first analyzed independently without harnessing model fits from previous studies, and are then contextualized with public data sets, requiring time-consuming data wrangling. We address these issues with sfaira, a single-cell data zoo for public data sets paired with a model zoo for executable pre-trained models. The data zoo is designed to facilitate contribution of data sets using ontologies for metadata. We propose an adaption of cross-entropy loss for cell type classification tailored to datasets annotated at different levels of coarseness. We demonstrate the utility of sfaira by training models across anatomic data partitions on 8 million cells.

Universal Gene Correction Approaches for ß-hemoglobinopathies Using CRISPR-Cas9 and Adeno-Associated Virus Serotype 6 Donor Templates.

Mutations in the human ß-globin gene are the cause of ß-hemoglobinopathies, one of the most common inherited single-gene blood disorders in the world. Novel therapeutic approaches are based on lentiviral vectors (LVs) or CRISPR-Cas9-mediated gene disruption to express adult hemoglobin (HbA), or to reactivate the completely functional fetal hemoglobin, respectively. Nonetheless, LVs present a risk of insertional mutagenesis, while gene-disrupting transcription factors (BCL11A, KLF1) involved in the fetal-to-adult hemoglobin switch might generate dysregulation of other cellular processes. Therefore, universal gene addition/correction approaches combining CRISPR-Cas9 and homology directed repair (HDR) by delivering a DNA repair template through adeno-associated virus could mitigate the limitations of both lentiviral gene transfer and gene disruption strategies, ensuring targeted integration and controlled transgene expression. In this study, we attained high rates of gene addition (up to 12%) and gene correction (up to 38%) in hematopoietic stem and progenitor cells from healthy donors without any cell sorting/enrichment or the application of HDR enhancers. Furthermore, these approaches were tested in heterozygous (ß0/ß+) and homozygous (ß0/ß0, ß+/ß+) ß-thalassemia patients, achieving a significant increase in HbA and demonstrating the universal therapeutic potential of this study for the treatment of ß-hemoglobinopathies.

Comparative targeting analysis of KLF1, BCL11A, and HBG1/2 in CD34+ HSPCs by CRISPR/Cas9 for the induction of fetal hemoglobin.

ß-hemoglobinopathies are caused by abnormal or absent production of hemoglobin in the blood due to mutations in the ß-globin gene (HBB). Imbalanced expression of adult hemoglobin (HbA) induces strong anemia in patients suffering from the disease. However, individuals with natural-occurring mutations in the HBB cluster or related genes, compensate this disparity through γ-globin expression and subsequent fetal hemoglobin (HbF) production. Several preclinical and clinical studies have been performed in order to induce HbF by knocking-down genes involved in HbF repression (KLF1 and BCL11A) or disrupting the binding sites of several transcription factors in the γ-globin gene (HBG1/2). In this study, we thoroughly compared the different CRISPR/Cas9 gene-disruption strategies by gene editing analysis and assessed their safety profile by RNA-seq and GUIDE-seq. All approaches reached therapeutic levels of HbF after gene editing and showed similar gene expression to the control sample, while no significant off-targets were detected by GUIDE-seq. Likewise, all three gene editing platforms were established in the GMP-grade CliniMACS Prodigy, achieving similar outcome to preclinical devices. Based on this gene editing comparative analysis, we concluded that BCL11A is the most clinically relevant approach while HBG1/2 could represent a promising alternative for the treatment of ß-hemoglobinopathies.

MHCquant: Automated and Reproducible Data Analysis for Immunopeptidomics.

Personalized multipeptide vaccines are currently being discussed intensively for tumor immunotherapy. In order to identify epitopes-short, immunogenic peptides-suitable for eliciting a tumor-specific immune response, human leukocyte antigen-presented peptides are isolated by immunoaffinity purification from cancer tissue samples and analyzed by liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS). Here, we present MHCquant, a fully automated, portable computational pipeline able to process LC-MS/MS data automatically and generate annotated, false discovery rate-controlled lists of (neo-)epitopes with associated relative quantification information. We could show that MHCquant achieves higher sensitivity than established methods. While obtaining the highest number of unique peptides, the rate of predicted MHC binders remains still comparable to other tools. Reprocessing of the data from a previously published study resulted in the identification of several neoepitopes not detected by previously applied methods. MHCquant integrates tailor-made pipeline components with existing open-source software into a coherent processing workflow. Container-based virtualization permits execution of this workflow without complex software installation, execution on cluster/cloud infrastructures, and full reproducibility of the results. Integration with the data analysis workbench KNIME enables easy mining of large-scale immunopeptidomics data sets. MHCquant is available as open-source software along with accompanying documentation on our website at https://www.openms.de/mhcquant/ .