Detection of early seeding of Richter transformation in chronic lymphocytic leukemia: scRNA-seq analysis
Richter transformation (RT) is a paradigmatic evolution of chronic lymphocytic leukemia (CLL) into a very aggressive large B cell lymphoma conferring a dismal prognosis. The mechanisms driving RT remain largely unknown. We characterized the whole genome, epigenome and transcriptome, combined with single-cell DNA/RNA-sequencing analyses and functional experiments, of 19 cases of CLL developing RT. Studying 54 longitudinal samples covering up to 19 years of disease course, we uncovered minute subclones carrying genomic, immunogenetic and transcriptomic features of RT cells already at CLL diagnosis, which were dormant for up to 19 years before transformation. We also identified new driver alterations, discovered a new mutational signature (SBS-RT), recognized an oxidative phosphorylation (OXPHOS)high–B cell receptor (BCR)low-signaling transcriptional axis in RT and showed that OXPHOS inhibition reduces the proliferation of RT cells. These findings demonstrate the early seed- ing of subclones driving advanced stages of cancer evolution and uncover potential therapeutic targets for RT.
This repository contains all the scripts, notebooks and reports to reproduce the scRNA-seq analysis of our paper "Early seeding of future Richter transformation in chronic lymphocytic leukemia", published in Nature Medicine in 2022. Here, we describe how to access the data, document the most important packages and versions used, and explain how to navigate the directories and files in this repository.
We have deposited the expression matrices, Seurat objects and metadata in Zenodo. You can download them following this link: https://doi.org/10.5281/zenodo.6631966, or executing the following commands in the command line:
wget https://zenodo.org/record/6631966/files/Nadeu2022_scRNAseq.zip
unzip Nadeu2022_scRNAseq.zip
cd Nadeu2022_scRNAseq
The "Nadeu2022_scRNAseq" data directory comes with its own README.md, which give a full description of the dataset. To ease things, we also provide it here:
The scRNA-seq data for this project was obtained for different patients (ids 12, 19, 63, 365, 3299) that underwent chronic lymphoyctic leukemia to Richter transformation. For each patient we have sequential clinical samples: disease, progression, treatment, Richter. The scRNA-seq data was sequenced in three different phases, which correspond to the three subprojects one can find in the associated metadata:
Smart-seq2 data for case 63. This was a proof-of-concept we carried out early on for this project, to assess if we could find Richter seeds in the initial time-points of the disease. Expression matrices were called using zUMIs.
As the proof-of-concept worked well, we scaled the sequencing to all samples from the other four cases: 12, 19, 365 and 3299. This corresponds to 10X Chromium v3 data, and was mapped with cellranger v4. For each case, we multiplexed the samples coming from different disease timepoints with cell hashing. This allowed us to remove batch effects, detect doublets and decrease sequencing costs.
Because for some samples we obtained very few cells in the previous phase (low starting material), we performed additionally 10X scRNA-seq for those samples: 365_07, 05_662, 00_30, 15-194, and 04-67.
The file "richter_metadata.csv" has all the relevant information (subproject, gem_id, library_id, library_name, type, and donor_id) for each sequencing library. Sequencing libraries are derived from individual "GEM wells". As defined by 10X Genomcis: "GEM wells are a set of partitioned cells (Gel Beads-in-emulsion) from a single 10x Genomics Chromium™ Chip channel. One or more sequencing libraries can be derived from a GEM well". More specifically, a GEM is: "an emulsion that contains a mixture of biochemistry reagents (uniquely barcoded gel beads) and zero, one, or more suspended cells/nuclei".
For cell-hashed libraries, each GEM well gives rise to two sequencing libraries: cDNA (reverse-trascribed mRNAs) and hasthags oligonucleotides (HTO). We give each GEM well a unique identifier, following the guideline described in this paper.
All the information in this file helped us locate the right fastq files in our high performance computer (HPC, cluster).
For cell-hashed libraries, we need to know which sample is identified by each barcode. This can be found in the "richter_feature_reference.csv" file, and we used it as input for the "Feature Reference CSV", which needs to be provided to cellranger as specificed in the Antibody Capture analysis vignette.
The file "sample_metadata.csv" contains all the metadata for each sample. Each sample represents a cell suspension obtained from a given disease time-point (i.e. diagnosis, progression, treatment, richter) for a given case (12, 19, 63, 365, 3299). The metadata includes donor_id, sample_id, tissue, time_point, treatment, etc.
See below.
For BCLLATLAS_10 and BCLLATLAS_29, we include the relevant files from the "outs" folder generated by cellranger:
- web_summary.html: html report with quality control metrics for the sequencing and mapping.
- metrics_summary.csv: csv file with the aforementioned QC metrics.
- filtered_feature_bc_matrix: a folder with 3 files, which correspond to the expresion matrix itself, the features and the cell barcodes.
For CLL_52, the expression matrix was called using zUMIs. The output is a bit convoluted, so we have simplified it and save the expression matrix as a csv file. In addition, the associated metadata can be found in the "CLL_52_cell_metadata.csv".
In the Zenodo repository we have deposited all the initial, intermediate and final Seurat Objects as .rds fies. Each object is the output of one of the analysis notebooks (.Rmd) that one can find at the GitHub repository associated with this publication (https://github.com/massonix/richter_transformation). We believe that this is very helpful, as one can start exploring the data at any step of interest that we show in our analysis.
In addition, we provide the output for the gene set enrichment analysis (GSEA) that we obtained with clusterProfiler. The file is called "patient_specific_gsea_raw_rt_vs_cll.rds" and before reading it one should load clusterProfiler.
These are the versions of the most important packages used throughout all the analysis:
CRAN:
- Seurat 4.0.3
- dplyr 1.0.6
- purrr 0.3.4
- stringr 1.4.0
- tidyr 1.1.3
- ggplot2 3.3.3
- patchwork 1.1.1
- harmony 1.0
- UpSetR 1.4.0
BioConductor:
You can check the versions of other packages at the "Session Information" section of each html report.
This repository contains 7 different analysis directories:
- 1-cellranger_mapping: scripts used to run cellranger in our cluster. It also contains QC metrics for different sequencing runs.
- 2-QC: quality control notebooks to assess the quality of sequencing and mapping, detect doublets, demultiplex hashtags, and filter poor-quality cells and genes.
- 3-clustering_and_annotation: we initially integrated all samples from all patients with Harmony, which allowed us to separate microenvironment from tumoral cells. Focusing on the latter, we then clustered and annotated all cells from each patient separately.
- 4-analysis_smart_seq2: notebooks with the full analysis of the Smart-seq2 data for patient 63.
- 5-clonal_evolution: notebooks that define the RT seed cells, and ensure that they are not artifacts.
- 6-differential_expression_analysis: differential expression analysis and gene set enrichment analysis between CLL and RT.
- 7-revision: notebooks to infer CNV and validate our findings with an external dataset.
The folder "bin" contains functions and variables that are used throughout different scripts. Finally, the folder "figures_and_tables" contains the scripts to generate the different figures derived from these analysis.
If you want to learn more about Richter Syndrome I suggest reading the following reviews: