Acute Myeloid Leukaemia (AML) stands as the predominant form of acute leukaemia affecting adults, presenting substantial challenges in patient management due to frequent disease relapse and the emergence of drug resistance. This study sets out to dissect the complexity of AML, using a rich bimodal dataset of approximately 300,000 bone marrow mononuclear cells (BMMCs) sourced from both AML patients and healthy controls. The data, generated via the 10x Genomics multiome platform, which pairs single nuclei ATAC and RNA sequencing, offers a multi-omics perspective on the tumour microenvironment and its cellular heterogeneity.
To navigate this complexity, we adopted a sequential integration strategy, creating a mini-atlas of BMMC cells. After meticulous rounds of quality control, our process began with the merging of data within the same modality across diverse donors, addressing batch effects, and integrating both transcriptome and chromatin accessibility modalities for clustering.
The outcome of this study is a precise and well-crafted integrated reference atlas of BMMC cells, based on profiling paired modalities in each cell, boosting the clustering resolution and improving cell type classification and annotation. The presented high-resolution bimodal resource exposes both intra- and inter-donor heterogeneity, serving as a vital asset for researchers and contributing to the existing pool of knowledge to bridge and annotate scATAC-seq and sn-RNA-seq single-modal datasets.
Significantly, this study showcases the first multiome analysis of AML samples to date, delivering profound insights into the epigenetic and transcriptional dynamics within leukaemia cells. In doing so, it provides deeper insights into the molecular mechanisms leading to drug resistance and paves the way for the development of targeted therapies. The ramifications of these findings are far-reaching, holding the promise of better patient outcomes and equipping the scientific community with an invaluable reference dataset.