FPKM and DESEq - RNAseq

Question

[carolgalah]

Hi..

I'm new to RNAseq and bioinformatics...

I aligned my samples with STAR, after aligning.... I got the value of FPKM through the cufflinks. I have 2 groups (treated and untreated) each group has 4 samples. I have FPKM value for each sample...

Can I get the genes differentially expressed between the groups with the FPKM values using the DEseq2? through R and with FPKM values, I can generate heatmap and PCA.

Can anyone help me by showing me some example script or indicating a tutorial? Please, I need to generate differentially expressed genes, heatmap, and PCA... using FPKM.

Thanks for any help!

Answer 1

[swbarnes2]

Can I get the genes differentially expressed between the groups with the FPKM values using the DEseq2?

No. DESeq needs raw gene counts.

[Kevin Blighe]

Also:

1. Cufflinks is old - the upgraded version is called StringTie.
2. Data measured by FPKM expression units is not suitable for differential expression comparisons
3. if you align your data with STAR, just use featureCounts to derive raw counts, which you will then use as input to DESeq2

For PCA:

• A: PCA plot from read count matrix from RNA-Seq
• PCAtools

[swbarnes2]

STAR also has a built-in gene counter, it seems to return much the same results as featureCounts, In my limited experience, so long as all you want is gene counts. And it helpfully returns results with all three strandedness options at once, so you can eyeball and make sure you are using the right one.

[carolgalah]

Thanks for your reply!

The STAR returns me the following files: BAM and reads by genes .... I get the featurecounts of the BAM files, correct? After generating the featurecounts, I run DESEq, as well as PCA.

[Kevin Blighe]

That is one of the many possible workflows, yes.

Answer 2

[dz2353]

Hi, As swbarnes said, you need to use reads count matrix as an input file. Before you use DESeq2 to analyze your data, you need to do some filter work in order to delete low expressed genes. After doing DESeq analysis, you will get a normalized gene expression matrix, and then you will do PCA analysis or drawing a volcano plot.

[GenoMax]

https://support.bioconductor.org/p/79855/