Pseudocount
  • Posts
  • About

Pseudocount

Bioinformatics statistics, checked by simulation: the pitfalls in RNA-seq, single-cell and omics analysis in R and Python, measured rather than asserted.

Omics statistics, measured. Each post takes one decision in an RNA-seq, single-cell or omics analysis, simulates data where the right answer is known, and shows what the decision does to the result. Code in R and Python, and every number on the page comes from code you can run.

qPCR statistics: test delta Ct, not fold change

qPCR statistics for the delta delta Ct method: run the t-test on delta Ct values, not on 2^-ddCt fold changes, and never count technical wells as replicates.

18 Nov 2026
28 min

GO enrichment: your background gene set decides

Which background (universe) should GO enrichment use in clusterProfiler or limma? Simulated data show the whole genome invents terms; use the genes you tested.

17 Nov 2026
25 min

Pseudobulk by sum or by mean, and how many cells?

Pseudobulk for single-cell DE: should you sum or average counts, and do donors with few cells need a floor? A simulation with edgeR, DESeq2 and limma answers.

16 Nov 2026
23 min

Seurat DE: integrated, SCT or RNA assay?

Which Seurat assay should differential expression use after integration or SCTransform? A simulation shows the integrated assay shrinks condition effects.

3 Nov 2026
16 min

Comparing cell type proportions in single-cell data

How to compare cell type proportions between conditions in single-cell data: a simulation of pooled chi-square and cell shuffling against propeller and edgeR.

2 Nov 2026
30 min

Samples nested in donors: sums or duplicateCorrelation

Several RNA-seq samples per donor, donors nested in groups: why independent-sample tests inflate the FDR, and how donor sums or duplicateCorrelation handle it.

1 Nov 2026
26 min

Paired samples in DESeq2: what ignoring patient costs

Paired design in DESeq2, edgeR and limma: what dropping patient from the design costs in power, whether it inflates FDR, and how to fix the not full rank error.

1 Oct 2026
22 min

Can you use TPM in DESeq2 or limma for DE analysis?

Can you use TPM in DESeq2 or a t-test for differential expression? A simulation measures what TPM input costs in power and false discoveries, and what to use.

29 Sept 2026
23 min

DESeq2 log2 fold change cutoff vs lfcThreshold

Is padj < 0.05 plus a log2 fold change cutoff a 2-fold result at 5% FDR? A simulation of the hidden false calls, and what lfcThreshold, glmTreat and treat cost.

27 Sept 2026
31 min

Poisson or negative binomial for RNA-seq counts?

Why a Poisson GLM calls false positives on RNA-seq counts with biological replicates, how to spot overdispersion, and what a negative binomial model fixes.

25 Sept 2026
22 min

Why correlations of relative abundances mislead

Proportions of independent taxa or cell types still correlate. A simulation of spurious correlation in relative abundance data and what CLR and rho can fix.

22 Sept 2026
25 min

Double dipping: marker gene p-values after clustering

Why are marker gene p-values after clustering so small? A simulation of double dipping in scRNA-seq, and what sample splitting and count splitting do about it.

18 Sept 2026
26 min

Interaction terms in DESeq2: what they test

How to read and test a DESeq2 interaction term, why significant in one genotype but not the other is not a difference, and how to get each genotype’s effect.

15 Sept 2026
23 min

Independent filtering: when gene filters break the FDR

Can you filter genes before multiple testing correction? Independent filtering on mean count keeps the FDR and adds power; a fold-change pre-filter breaks it.

11 Sept 2026
22 min

How to read a p-value histogram in RNA-seq

How to read a p-value histogram from an RNA-seq differential expression analysis: the healthy, anti-conservative, conservative and U shapes, and each fix.

8 Sept 2026
18 min

CPM vs TMM: composition bias in RNA-seq counts

Why CPM makes unchanged genes look down when a few genes take a big share of reads, and how TMM and DESeq2 median-of-ratios fix composition bias in RNA-seq.

4 Sept 2026
25 min

Batch confounded with condition: can correction help?

Batch confounded with condition in RNA-seq: when a batch term in the design can rescue the analysis, when it cannot, and why removeBatchEffect before DE fails.

1 Sept 2026
21 min

Pseudobulk vs cell-level DE tests in scRNA-seq

Is it fine to treat cells as replicates in single-cell differential expression? A simulation of Wilcoxon cell-level tests against pseudobulk edgeR and DESeq2.

28 Aug 2026
22 min

What log2(x + 1) does to low-count fold changes

Adding 1 before log2 shrinks the fold changes of low-count genes by an amount set by expression and by the constant. A simulation against DESeq2 and edgeR.

25 Aug 2026
22 min

Welcome to Pseudocount

What Pseudocount covers and how its bioinformatics statistics tutorials work: one analysis decision per post, tested on simulated omics data with a known answer.

24 Aug 2026
1 min
No matching items

© 2026 Pseudocount

 
  • Privacy

  • Terms

  • Disclaimer