Additional files for A practical guide to methods controlling false discoveries in computational biology

This repository contains the knitted Rmarkdown vignettes for simulations and case studies described in A practical guide to methods controlling false discoveries in computational biology. The Rmarkdown files are also available in the companion benchmark-fdr repo on GitHub.

Yeast in silico experiments

• Additional file 2 - Yeast in silico experiments I.
• Analysis and benchmarking results under the null, and using a unimodal alternative effect size distribution and large proportion (30%) of non-nulls using yeast RNA-seq data.
• Additional file 3 - Yeast in silico experiments II.
• Analysis and benchmarking results using a unimodal alternative effect size distribution and small proportion (7.5%) of non-nulls using yeast RNA-seq data.
• Additional file 4 - Yeast in silico experiments III.
• Analysis and benchmarking results using a bimodal alternative effect size distribution and large proportion (30%) of non-nulls using yeast RNA-seq data.
• Additional file 5 - Yeast in silico experiments IV.
• Analysis and benchmarking results using a bimodal alternative effect size distribution and small proportion (7.5%) of non-nulls using yeast RNA-seq data.

Polyester in silico experiments

• Additional file 6 - Polyester in silico experiments.
• Analysis and benchmarking results using a bimodal alternative effect size distribution and large proportion (30%) of non-nulls using RNA-seq data simulated using Polyester.

Simulation experiments

• Additional file 7 - Simulations I: Null.
• Analysis and benchmarking results of simulation settings with only null tests, using normal, t, and chi-squared distributed test statistics.
• Additional file 8 - Simulations II: Informative (cubic).
• Analysis and benchmarking results of simulation settings with cubic informative covariate and normal, t, and chi-squared distributed test statistics.
• Additional file 9 - Simulations III: Informative (step).
• Analysis and benchmarking results of simulation settings with step informative covariate and normal, t, and chi-squared distributed test statistics.
• Additional file 10 - Simulations IV: Informative (sine).
• Analysis and benchmarking results of simulation settings with sine informative covariate and normal, t, and chi-squared distributed test statistics.
• Additional file 11 - Simulations V: Informative (cosine).
• Analysis and benchmarking results of simulation settings with cosine informative covariate and normal, t, and chi-squared distributed test statistics.
• Additional file 12 - Simulations VI: Unimodal Effect Sizes.
• Analysis and benchmarking results of simulation settings with cubic informative covariate, normal test statistics and unimodal effect size distributions.
• Additional file 13 - Simulations VII: Unimodal Effect Sizes (\(t_{11}\) test statistics).
• Analysis and benchmarking results of simulation settings with cubic informative covariate, \(t_{11}\) distributed test statistics and unimodal effect size distributions.
• Additional file 14 - Simulations VIII: Unimodal Effect Sizes (25% non-null).
• Analysis and benchmarking results of simulation settings with cubic informative covariate, normal test statistics, unimodal effect size distributions, and higher (25% vs 10%) proportion of non-null tests.
• Additional file 15 - Simulations IX: Varying \(M\) Tests.
• Analysis and benchmarking results of simulation settings with sine informative covariate, normal test statistics, and varying total number of hypothesis tests.
• Additional file 16 - Simulations X: Varying Null Proportion.
• Analysis and benchmarking results of simulation settings with sine informative covariate, normal test statistics, and varying proportion of null hypotheses.
• Additional file 17 - Simulations XI: Varying Null Proportion (\(t_{11}\) test statistics).
• Analysis and benchmarking results of simulation settings with sine informative covariate, \(t_{11}\) distributed test statistics, and varying proportion of null hypotheses.
• Additional file 18 - Simulations XII: Varying Informativeness (continuous \(p(x;\delta)\)).
• Analysis and benchmarking results of simulation settings with informative covariates of varying informativeness using a continuous relationship between the covariate and the null proportion.
• Additional file 19 - Simulations XIII: Varying Informativeness (discrete \(p(x;\delta)\)).
• Analysis and benchmarking results of simulation settings with informative covariates of varying informativeness using a discrete relationship between the covariate and the null proportion.
• Additional file 20 - Simulations XIV: AdaPT with null option.
• Analysis and benchmarking results of simulation settings with step informative covariate comparing AdaPT with and without a null model option.

Case studies

• Additional file 21 - Case study: Genome-wide association studies.
• Analysis and benchmarking results of a meta-analysis of genome-wide association studies of body mass index.
• Additional file 22 - Case study: Gene set analyses I.
• Gene set analysis of differentially expressed mouse genes after HSC differentiation using goSeq.
• Additional file 23 - Case study: Gene set analyses II.
• Gene set analysis of differentially expressed human genes between cerebellum and cerebral cortex using goSeq.
• Additional file 24 - Case study: Gene set analyses III.
• Gene set analysis and benchmarking results of differentially expressed mouse genes after HSC differentiation using GSEA.
• Additional file 25 - Case study: Gene set analyses IV.
• Gene set analysis and benchmarking results of differentially expressed human genes between cerebellum and cerebral cortex using GSEA.
• Additional file 26 - Case study: Differential gene expression in bulk RNA-seq I.
• Differential expression analysis and benchmarking results of human genes between nucleus accumbens and putamen.
• Additional file 27 - Case study: Differential gene expression in bulk RNA-seq II.
• Differential expression analysis and benchmarking results of human genes between microRNA knockdown and control mouse cells.
• Additional file 28 - Case study: Differential gene expression in single-cell RNA-seq I.
• Single-cell differential expression analysis and benchmarking results between human glioblastoma tumor cells and nearby controls using MAST.
• Additional file 29 - Case study: Differential gene expression in single-cell RNA-seq II.
• Single-cell differential expression analysis and benchmarking results between human glioblastoma tumor cells and nearby controls using scDD.
• Additional file 30 - Case study: Differential gene expression in single-cell RNA-seq III.
• Single-cell differential expression analysis and benchmarking results between human glioblastoma tumor cells and nearby controls using Wilcox.
• Additional file 31 - Case study: Differential gene expression in single-cell RNA-seq IV.
• Single-cell differential expression analysis and benchmarking results between stimulated murine macrophage cells and controls using MAST.
• Additional file 32 - Case study: Differential gene expression in single-cell RNA-seq V.
• Single-cell differential expression analysis and benchmarking results between stimulated murine macrophage cells and controls using scDD.
• Additional file 33 - Case study: Differential gene expression in single-cell RNA-seq VI.
• Single-cell differential expression analysis and benchmarking results between stimulated murine macrophage cells and controls using Wilcox.
• Additional file 34 - Case study: Differential binding in ChIP-seq I.
• Differential binding analysis and benchmarking results of H3K4me3 between two cell lines in promoter regions using DESeq2.
• Additional file 35 - Case study: Differential binding in ChIP-seq II.
• Differential binding analysis and benchmarking results of H3K4me3 between two cell lines using csaw.
• Additional file 36 - Case study: Differential binding in ChIP-seq III.
• Differential binding analysis and benchmarking results of CREB-binding protein between knockout and wild-type mice using csaw.
• Additional file 37 - Case study: Differential abundance testing in microbiome data analysis I.
• Differential abundance analysis and benchmarking results of obesity.
• Additional file 38 - Case study: Differential abundance testing in microbiome data analysis II.
• Differential abundance analysis and benchmarking results of IBD.
• Additional file 39 - Case study: Differential abundance testing in microbiome data analysis III.
• Differential abundance analysis and benchmarking results of infectious diarrhea.
• Additional file 40 - Case study: Differential abundance testing in microbiome data analysis IV.
• Differential abundance analysis and benchmarking results of colorectal cancer.
• Additional file 41 - Case study: Correlation testing in microbiome data analysis.
• Correlation analysis and benchmarking results of wastewater contaminants.

Referee response

• Referee response R1 - Simulations: Violation of covariate independence under the null.
• Analysis and benchmarking results of simulation settings with either weak or strong dependence between p-values and covariate under the null (assumption violated), using normal distributed test statistics.