Hello everyone,

I am working on a halophyte plant species and performed RNA-seq differential expression analysis using Galaxy. I am facing an issue in my reference-based analysis, where many genes show large log2 fold changes (|log2FC| > 1), but none (or very few) have significant adjusted p-values (padj < 0.05), making it difficult to interpret results or generate a meaningful volcano plot.

Experimental design: two conditions (Control vs Salt stress), three biological replicates per condition

Reference-based workflow (Galaxy):

1. Quality control and trimming: FastQC, Trimmomatic, MultiQC

   1. Reads grouped into a collection (6 samples total)

   2. Alignment: HISAT2

   3. Transcript assembly: StringTie

   4. Read counting: HTSeq-count

   5. Differential expression: DESeq2

In the DESeq2 results, several genes show relatively high absolute log2FC values, but all adjusted p-values remain above 0.05.

In addition, I performed a de novo transcriptome-based analysis using the same RNA-seq data, which identified ~20 differentially expressed genes with significant adjusted p-values. I therefore expected at least a similar or improved level of significance in the reference-based approach, but this was not observed.

I would appreciate insights on why a reference-based RNA-seq analysis shows large log2 fold changes but no significant adjusted p-values, even though I pooled five plant samples per replicate in RNA extraction (3 replicates per condition), and a de novo analysis identified few significant DEGs.

As I am fairly new to bioinformatics, I would like to hear your opinions on the possible reasons for this and on measures or approaches that could help address it.

Thank you Chamara Lakshitha