Example:
dds <- DESeq2::DESeqDataSetFromMatrix(countData, colData, design=~condition)
dds <- DESeq2::estimateSizeFactors(dds)
normalized_counts <- DESeq2::counts(dds, normalized=TRUE)

This is the gold standard for between-sample DE analysis.

#statstab #266 When to use negative binomial and Poisson regression

Thoughts: A heated debate and condemnation of NB by @noah_greifer. Seems term confusion is also an issue in the field.

#poisson #regression #negativebinomial #countdata #overdispersion

stats.stackexchange.com/questions/65...

7/
New to RNA-seq? You'll run:
dds <- DESeqDataSetFromMatrix(countData, colData, design = ~ condition)
dds <- DESeq(dds)

But 10 years in, you pause:
Are the replicates balanced?

Any hidden batch effects?

Should I use lfcShrink() before plotting?

pubmed.ncbi.nlm.nih.gov/38564640/
#statsky #stats #methods #genomics #countData

You can see every line of R code run:

rb.codegen.enable()
results = rb.deseq2(counts_df, meta_df, design="~ batch + condition")
R> library(DESeq2)
R> dds <- DESeqDataSetFromMatrix(countData=counts, colData=metadata, design=~ batch + condition)
R> dds <- DESeq(dds)
R> res <- results(dds, alpha=0.05)

7/
New to RNA-seq? You'll run:
dds <- DESeqDataSetFromMatrix(countData, colData, design = ~ condition)
dds <- DESeq(dds)

But 10 years in, you pause:
Are the replicates balanced?

Any hidden batch effects?

Should I use lfcShrink() before plotting?

Example:
dds <- DESeq2::DESeqDataSetFromMatrix(countData, colData, design=~condition)
dds <- DESeq2::estimateSizeFactors(dds)
normalized_counts <- DESeq2::counts(dds, normalized=TRUE)

This is the gold standard for between-sample DE analysis.

7/
New to RNA-seq? You'll run:
dds <- DESeqDataSetFromMatrix(countData, colData, design = ~ condition)
dds <- DESeq(dds)

But 10 years in, you pause:
Are the replicates balanced?

Any hidden batch effects?

Should I use lfcShrink() before plotting?

Example:
dds <- DESeq2::DESeqDataSetFromMatrix(countData, colData, design=~condition)
dds <- DESeq2::estimateSizeFactors(dds)
normalized_counts <- DESeq2::counts(dds, normalized=TRUE)

This is the gold standard for between-sample DE analysis.

This #CrunchDAO thingy is frustrating me because I'm unfamiliar with using Python for single cell analysis. If I had the data in R, I'd be able to process and summarise it in an hour or so.

I want something that functions like tapply(countData, status, sum), and am currently doing a double for-loop