R scripts for the analyses in the DECENT manuscript

Scripts:

exam_dropout_model.Rmd

• Examine dropout model assumptions using ERCC spike-in data.
• Fig.1, sFig.1, sFig.2

overdisp_zeroinfl_pre_dropout.R

• Looking for overdispersion and zero-inflation in the pre-dropout count using scRNA-seq data based on the DECENT model.
• sFig.3, sFig.4

test_smfish.R

• Testing for overdispersion and zero-inflation in the smFISH data.
• sFig.5

misspecify_eta.R

• Plotting probability mass function of the observed data given various levels of misspecified eta.
• sFig.6

sim.R

• Simulation studies.
• Fig.2, Fig.3, sFig.7

tung.R

• Analyses of Tung et al. dataset.
• sFig.8

soumillon.R

• Analyses of Soumillon et al. dataset.

savas.R

• Analyses of Savas et al. dataset.

chen.R

• Analyses of Chen et al. dataset.

benchmark_plot.R

• Making benchmarking plots.
• Fig.4, Fig.5, Fig.6, sFig.9, sFig.10, sFig.11

func_de_methods.R

• Utility functions.

sim_rep.R

• Repeat simulation study 20 times.

r1_plots.R

• New plots for revision 1.

Input data:

data.savas.rds

• The UMI count matrix of the Savas et al. data.

ct.savas.rds

• A vector denoting the two cell types in the Savas et al. data.

simdata_ZINB_BB_Tung6_OD_kb.RData

• Simulated dataset.

simdata_ZINB_BB_Tung6_OD_kb_010119/

• Another 20 simulated datasets.

ercc_length.txt

• Lengths of ERCC spike-ins.

cms_095046.txt

• Information of ERCC spike-in mix.

• Other input data can be directly downloaded from public repositories as noted in the scripts.

Saved data:

Tung_benckmark.rda Soumillon_benckmark.rda Savas_benckmark.rda Chen_benckmark.rda fpr.plots.tung.rda fpr.plots.soumillon.rda

DECENT output:

sim/ tung/ tung_nb/ soumillon/ savas/ chen/ zeisel/ zeisel_nb/

TASC output:

tasc/