Data repository for "An in vitro model of tumor heterogeneity resolves genetic, epigenetic, and stochastic sources of cell state variability," Hayford et al. (2021), PLoS Biology 19 : e3000797; DOI: 10.1371/journal.pbio.3000797

---

*Instructions for creating panels in all main and supplementary figures based on experimental and simulated data in this repository

• MAIN FIGURES

  • FIGURE 1: N/A

  • FIGURE 2

    Panels A and C: In the DrugResponse directory, run DrugResponse.R, which pulls data from the two Parental-*.csv files in the directory and the well conditions in the DrugResponse/Platemaps subdirectory.

    Panels B and D: In the cFP directory, run cFP.R, which pulls data from the 10 cFP_*.csv files in the directory.

  • FIGURE 3

    Panel A: In the WES directory, run WES.R, which pulls data from mutations_byChromosome.csv.

    Panels B, C, and D: In the WES directory, run WES.R, which pulls data from the vep_*.txt files in the directory and uses the database in the RData object in RefCDS_human_GRCH38.p12.rda to cross-reference variants. NOTE: The vep_*.txt files must be manually unzipped before running WES.R.

    Panel E: In the scRNAseq/inferCNV subdirectory, run inferCNV.R, which pulls a counts matrix from the RData object in PC9.CLV.10x.counts.matrix.rds, included in the directory. Necessary annotation and gene order files are also provided.

    Panel F: In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories. Scripts to de-multiplex hashed raw data and outputs are included in the scRNAseq/HTO_identification subdirectory. A full matrix of de-multiplexed counts is included as PC9_scRNAseqCounts_HTOdemux.csv.zip.

    Panel G: In the GO directory, run GO_correlation.R, which pulls data from mutations_DEGs-hg38.RData, a file that compiles all IMPACT genetic mutations (from the WES directory) and differentially expressed genes (DEGs; from the scRNAseq directory).

  • FIGURE 4

    Panel A: In the WES folder, run WES.R, which pulls data from mutations_byChromosome.csv.

    Panels B, C, and D: In the WES directory, run WES.R, which pulls data from the vep_*.txt files in the directory and uses the database in the RData object in RefCDS_human_GRCH38.p12.rda to cross-reference variants.

    Panel E: In the scRNAseq/inferCNV subdirectory, run inferCNV.R, which pulls a counts matrix from the RData object in PC9.VUDS.10x.counts.matrix.rds (created in inferCNV.R). Necessary annotation and gene order files are also provided.

    Panel F: In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories. Scripts to de-multiplex hashed raw data and outputs are provided in the scRNAseq/HTO_identification subdirectory. A full matrix of de-multiplexed counts is included as PC9_scRNAseqCounts_HTOdemux.csv.zip.

    Panel G: In the GO folder, run GO_correlation.R, which pulls data from mutations_DEGs-hg38.RData, a file that compiles all IMPACT genetic mutations (from the WES directory) and differentially expressed genes (DEGs; from the scRNAseq directory).

  • FIGURE 5

    Panels A and E: In the cFP directory, run cFP.R, which pulls data from the trajectories_*.csv files in the directory.

    Panels B and F: In the cFP directory, run cFP.R, which pulls simulated data from the trajectories_*.csv files in the directory. Model trajectories are representative examples of a larger simulation scan (*.py models in the Simulations directory).

    Panels C and G: In the cFP directory, run cFP.R, which pulls simulated data from the distributions_*.csv files in the directory. Model distributions were calculated from example trajectories as part of a larger simulation scan (*.py models in the Simulations directory). For each subline, the mean and confidence interval reported on the plot is calculated based on 100 bootstrapped p-values provided in one of the ADbootstrap*.csv files.

    Panels D and H: In the Simulations directory, run plotParameterScan.R, which pulls data from the *_lowVal.csv files in the directory.

  • FIGURE 6: N/A

• SUPPLEMENTARY FIGURES

  • SUPPLEMENTARY FIGURE S1

    Panel A: Screenshot of the EGFR gene from the Integrative Genomics Viewer (IGV) based on raw exome sequencing data (available in the Sequence Read Archive (SRA) at accession #PRJNA632351). Image is stored as PC9-EGFRgene_mutations_ex19delCommon.svg in the WES directory.

    Panel B: N/A

  • SUPPLEMENTARY FIGURE S2

    Panels A, B, and C: In the cFP directory, run cFP.R, which pulls data from the trajectories_*.csv files in the directory. Data from overlays in panel C come from the PopD_trajectories.RData object.

  • SUPPLEMENTARY FIGURE S3

    Panel A: In the WES directory, run WES.R, which pulls data from number_mutations.csv in the directory.

    Panel B: In the WES directory, run WES.R, which pulls data from samples_called_vars_named.vcf.gz in the directory. Directions to download reference FASTA and GTF files are provided in WES.R.

    Panel C: In the WES directory, run WES.R, which pulls data from shared_variants_CLV.csv in the directory.

    Panel D: In the WES directory, run WES.R, which pulls data from shared_variants_sublines.csv in the directory.

    Panel E: In the WES directory, run WES.R, which pulls data from shared_variants_VUDSlines.csv in the directory.

  • SUPPLEMENTARY FIGURE S4

    Panels A and B: In the WES directory, run WES.R, which pulls data from the vep_*.txt files in the directory and uses the database in the RData object in RefCDS_human_GRCH38.p12.rda to cross-reference variants.

  • SUPPLEMENTARY FIGURE S5

    Panel A: Screenshot of the summarized output from the Cell Ranger quality control analysis on the scRNA-seq library (available in the Gene Expression Omnibus (GEO) data repository at accession #GSE150084). Settings are shown in the image, which is stored as CellRanger_PC9.svg in the scRNAseq directory.

    Panel B: In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories.

  • SUPPLEMENTARY FIGURE S6

    Panels A and B: In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories and subsets data by cell line versions.

    Panels C and D: In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories and subsets data by sublines.

    Panels E and F: In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories.

  • SUPPLEMENTARY FIGURE S7

    Panels A and B: In the RNAseq directory, run RNAseq.R, which pulls from all 8 *_featurecounts.txt files in the directory. These files were created using the Bash script in RNAseq_processing.txt. NOTE: The *_featurecounts.txt files must be manually unzipped before running RNAseq.R.

  • SUPPLEMENTARY FIGURE S8

    In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories. Input hallmark gene signature (.gmt) files can be found in the scRNAseq/VISION_gmt/hallmark subdirectory.

  • SUPPLEMENTARY FIGURE S9

    In the GO directory, run semanticSimilarity.R, which pulls data from mutations_DEGs-hg38.RData, a file that compiles all IMPACT genetic mutations (from the WES directory) and differentially expressed genes (DEGs; from the scRNAseq directory). Directions for downloading reference GTF file are provided in semanticSimilarity.R.

  • SUPPLEMENTARY FIGURE S10

    In the scRNAseq/inferCNV directory, run inferCNV.R, which pulls a counts matrix from the RData object in PC9.VUDS.10x.counts.matrix.rds (created in inferCNV.R). Necessary annotation and gene order files are also provided in the directory.

  • SUPPLEMENTARY FIGURE S11: N/A

  • SUPPLEMENTARY FIGURE S12

    Panel A: In the cFP directory, run cFP.R, which pulls data from the trajectories_*.csv files in the directory.

    Panel B: In the cFP directory, run cFP.R, which pulls data from the trajectories_*.csv files in the directory. Model trajectories are representative examples of a larger simulation scan (*.py models in the Simulations directory).

    Panel C: In the cFP directory, run cFP.R, which pulls simulated data from the distributions_*.csv files in the directory. Model distributions were calculated from example trajectories as part of a larger simulation scan (*.py models in the Simulations directory). For each subline, the mean and confidence interval reported on the plot is calculated based on 100 bootstrapped p-values provided in one of the ADbootstrap*.csv files.

    Panel D: In the Simulations directory, run plotParameterScan.R, which pulls from the *_lowVal.csv files in the directory.

  • SUPPLEMENTARY FIGURE S13

    Panels A and B: In the WES directory, run WES.R, which pulls data from samples_called_vars_named.vcf.gz in the directory.

  • SUPPLEMENTARY FIGURE S14

    In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories.

  • SUPPLEMENTARY FIGURE S15

    In the scRNAseq directory, run scRNAseq.R, which pulls from 10x Genomics reduced data in the scRNAseq/read_count and scRNAseq/umi_count subdirectories.