GVP_2023: Studying stochastic systems biology of the cell with single-cell genomics data

This repository contains all the scripts necessary to reproduce figures in "Studying stochastic systems biology of the cell with single-cell genomics data." All figure components are output in figs/ in pdf and png (450 DPI) formats. Font sizes vary.

• Figure 1
  • Panel d: run gg230227_cartoons.ipynb. This will generate the files fig1d_txmodels (transcription models) and fig1d_reactors (reactor models).
• Figure 2
  • Panel a: run the first cell and the last two cells of gg230117_emptydrops.ipynb. This will generate the file fig2a_poissprod.
  • Panels b, c, d, e: download the raw data from Zenodo. Run gg230117_emptydrops.ipynb. This will generate the files fig2bcde, which has the content for the body, and emptydrops_supp_pbmc_1k_v3, which has the content for the supplement.
  • Supplements: run the five other emptydrops notebooks. They will generate emptydrops_supp_brain_nuc_5k_v3, emptydrops_supp_desai_dmso, emptydrops_supp_heart_1k_v3, emptydrops_supp_neuron_1k_v3, and emptydrops_supp_pbmc_1k_v2.
    • Note: each notebook can take up to 15 minutes to run (gene-gene correlations are slow to calculate and store).
• Figure 3
  • Panel a: run gg230227_cartoons.ipynb. This will generate the files fig3a_gou, fig3a_cir, and fig3a_tele.
  • Panel b: run gg230221_modelfit.ipynb until the heading "AIC". This will generate the file fig3b.
  • Panel c: run gg230221_modelfit.ipynb from "AIC" to the end. This will generate the file fig3c.
• Figure 4
  • Panel a:
    • Run gg230227_cartoons.ipynb. This will generate the files fig4a_pfr_iad, fig4a_cstr_iad, and fig4a_lfr_iad.
    • Run gg230302_reactors.ipynb up to the heading "Visualization". This will generate the files fig4a_burstsize_cartoon, fig4a_copynumber_traj, and fig4a_counts_over_time.
  • Panel b:
    • Continue running gg230302_reactors.ipynb from "Visualization" to "Cell number dependence". This will generate the file fig4b.
  • Panel c:
    • Continue running gg230302_reactors.ipynb from "Cell number dependence" to "Parameter value dependence". This will generate the file fig4c.
  • Panels d, e: Continue running gg230302_reactors.ipynb after "Parameter value dependence". This will generate the files fig4d and fig4e.
• Figure 5
  • Panel b: run gg230131_technoisebern.ipynb. This will generate the file fig5b.
  • Panels c, d: download the raw data from Zenodo, specifically GP_2021_3_fits.tar.gz, which contains the search data and result files. Install Monod. Run gg230203_monod_technoise.ipynb. This will generate the files fig5c and fig5d.

Notes:

• I have tried to make the notebooks "safe," i.e., running them should take a reasonable amount of time, not perform any new computations, and not overwrite any data.
  • This is not the case for the empty droplet code, which is slow and has to redo all of the computations from zero.
  • Intermediate data are stored in data/. Note, however, that not every single file in that directory is used.