Human dermal fibroblast clonality project

Project overview

This project investigates clonality in human dermal fibroblast cell populations in 32 cell lines from distinct donors, using bulk whole-exome sequencing and single-cell RNA-sequencing data.

Key findings:

• A novel approach for integrating DNA-seq and single-cell RNA-seq data to reconstruct clonal substructure and single-cell transcriptomes.
• A new computational method, cardelino, to map single-cell RNA-seq profiles to clones.
• Evidence for non-neutral evolution of clonal populations in human fibroblasts.
• Proliferation and cell cycle pathways are commonly distorted in mutated clonal populations, with implications for cancer and ageing.

For a richer overview, see the About page.

Data pre-processing

The data pre-processing for this project from the raw data described above is complicated and computationally expensive, so this repository does not reproduce the data pre-processing in an automated way. However, we provide the source code for the Snakemake workflow for data pre-processing in this repository. Docker images providing the computing environment and software used are publicly available, split into an image for command line bioinformatics tools and an R installation with necessary packages installed.

If you would like to pre-process the data from raw reads to results as we have, please consult our description of how to run the workflow.

Analyses

Here we present the reproducible the results of our analyses. They were generated by rendering the R Markdown documents into webpages available at the links below.

The results presented in the paper were produced with these analyses.

1. Simulation results.

2. Overview of lines

3. Selection models

4. Analysis of clonal prevalences.

5. Analysis for the example cell line joxm.

6. Variance components analysis.

7. Differential expression analysis.

8. Analysis of effects of somatic variants on cis gene expression.

Data availability

This is a complicated project, and reproducing all of the results presented, especially from raw data is highly non-trivial. Nevertheless, we have made all data available so that everything is entirely reproducible.

Single-cell RNA-seq data have been deposited in the ArrayExpress database at EMBL-EBI under accession number E-MTAB-7167. Whole-exome sequencing data is available through the HipSci portal. Processed data and large results files are available from Zenodo with DOI 10.5281/zenodo.1403510.

To set up the project to reproduce our analyses, first clone the source code repository from GitHub. Next, download all of the reference, metadata and results files and add them to the (cloned) project folder with the following structure:

.
├── data
│   ├── canopy
│   │   ├── canopy_results.*.rds
│   ├── cell_assignment
│   │   ├── cardelino_results.*.rds
│   ├── de_analysis_FTv62
│   │   ├── cellcycle_analyses
│   │   │   ├── filt_lenient.all_filt_sites.de_results_unstimulated_cells.cc.rds
│   │   │   ├── filt_lenient.cell_coverage_sites.de_results_unstimulated_cells.cc.rds
│   │   │   ├── filt_strict.all_filt_sites.de_results_unstimulated_cells.cc.rds
│   │   │   └── filt_strict.cell_coverage_sites.de_results_unstimulated_cells.cc.rds
│   │   ├── filt_lenient.all_filt_sites.de_results_unstimulated_cells.rds
│   │   ├── filt_lenient.cell_coverage_sites.de_results_unstimulated_cells.rds
│   │   ├── filt_strict.all_filt_sites.de_results_unstimulated_cells.rds
│   │   └── filt_strict.cell_coverage_sites.de_results_unstimulated_cells.rds
│   ├── donor_info_070818.txt
│   ├── donor_info_core.csv
│   ├── donor_neutrality.tsv
│   ├── exome-point-mutations
│   │   ├── high-vs-low-exomes.v62.ft.alldonors-filt_lenient.all_filt_sites.vep_most_severe_csq.txt
│   │   └── high-vs-low-exomes.v62.ft.filt_lenient-alldonors.txt.gz
│   ├── human_H_v5p2.rdata
│   ├── human_c2_v5p2.rdata
│   ├── human_c6_v5p2.rdata
│   ├── neg-bin-rsquared-petr.csv
│   ├── neutralitytestr-petr.tsv
|   ├── sces
│   │   ├── sce_*.rds
│   ├── selection
│   │   ├── neg-bin-params-fit.csv
│   │   ├── neg-bin-rsquared-fit.csv
│   ├── simulations
│   │   ├── *.filt_lenient.cell_coverage_sites.mult.rds
│   │   ├── *.simulate.rds
│   └── variance_components
│       ├── covar_all.csv
│       ├── donorVar
│       │   ├── *.var_part.var1.csv
│       ├── fit_all_gene_highVar.csv
│       ├── fit_per_gene_highVar.csv
│       ├── gene_info_all.csv
│       └── logcnt_all.csv
├── metadata
│   ├── cell_metadata.csv
│   └── data_processing_metadata.tsv
├── references
│   ├── 1000G_phase1.indels.hg19.sites.vcf.gz
│   ├── GRCh37.p13.genome.ERCC92.fa
│   ├── Homo_sapiens.GRCh37.rel75.cdna.all.ERCC92.fa.gz
│   ├── Mills_and_1000G_gold_standard.indels.hg19.sites.vcf.gz
│   ├── dbsnp_138.hg19.biallelicSNPs.HumanCoreExome12.Top1000ExpressedIpsGenes.Maf0.01.HWE0.0001.HipSci.vcf.gz
│   ├── dbsnp_138.hg19.vcf.gz
│   ├── gencode.v19.annotation_ERCC.gtf
│   ├── hipsci.wec.gtarray.HumanCoreExome.imputed_phased.20170327.genotypes.allchr.fibro_samples_v2_filt_vars_sorted_oa.vcf.gz
│   ├── hipsci.wec.gtarray.HumanCoreExome.imputed_phased.20170327.genotypes.allchr.fibro_samples_v2_filt_vars_sorted_oa.vcf.gz.csi
│   └── knownIndels.intervals

For simplicity, we ignore all the directories and files present in the source code repository (that you should have clones) to focus just on where you should add the files downloaded from Zenodo. Yes, it’s still complicated, but such is life.

There is a large number of canopy_results.*.rds files: these should be stored in the data/canopy directory. Similarly, all of the cardelino_results.*.rds files should be stored in data/cell_assignment. All of the SingleCellExperiment object files (sce_*.rds) should be stored in data/sces. Simulation results files (*.mult.rds; *.simulate.rds) should be stored in data/simulations. Variance components results should be stored in data/variance_components as shown above.

Differential expression results belong in data/de_analysis_FTv62.

Metadata files belong in metadata. Reference files belong in references.

With the data downloaded and organised as above, you will be able to reproduce the analyses presented in the RMarkdown files linked to above and, if desired, even run the whole analysis pipeline from raw reads to results following these instructions.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.