Hello, everyone
I am analyzing a datasets of RNA-seq from different stages of cervical cancer. I would like to identify up or down-regulated immune related genes between the different stage. I have a list of the immune related genes and I would like to filter these genes before the differential expression analysis, using Deseq2, but I don't know if this could be a correct method to achieve my objective. Any help or suggestion? Thank you so much.
Jonathan Pena
•
written
by
jonathanpa12 •
1 answer
If there is a certain subset of genes you are interested in, I would do the filtering after applying the DESeq2 analysis. This is because information across all genes is used for the calculation of normalisation factors and variances. If you are worried that filtering after means you loose power due to multiple testing, you can run DESeq2, subset to the genes you are interested in, and then re-calculate the padj column using p.adjust. I would also probably turn off filtering for low expression.
Log in to answer this question.
-
differential gene expression in male and female samples
written by BioS •
-
log2Fold Change as input for k-means analysis
written by concetta •
-
expression trend during cancer progression
written by Amir.Asgary •
-
LncRNA and Immune-related genes
written by jonathanpa12 •
-
Meta analysis to find candidate genes
written by au.rinki.bio •
-
RNA seq query
written by komal •
-
changing the pattern of expression in two sequential stages of cancer
written by nazaninhoseinkhan
-
Copy number variation between different cancer stages
written by nazaninhoseinkhan
-
Same KEGG pathways enriched by up- and down-regulated genes
written by mahmeh •
-
Patterns within groups with RNA-seq
written by gaiusjaugustus
Why do you want to filter? Is the reason that those immune-related genes do not show up in the final list of differentially expressed genes? Or do you have a very strong biological rationale that the immune-related genes should be the ones driving the differences between the different conditions that you're analyzing?
The final list of differentially of expressed gene shows only two immune-related gene. I was thinking that those genes usually have very low expression rates but biologically minimal differences in gene expression sometimes may produce significant changes in immune response, Because of this I thought maybe filtering them before differential expression analysis could be an option, but I don’t have a lot of experience in this type of analysis and I don’t know if it would be right to do it.
When you say "final list of differentially of expressed gene shows only two immune-related gene", do you mean they are not DE, or they are NA?
Sorry, I must learn to ask better. I used Deseq2 to make the differential expression analysis and I got 11 genes using as significance thresholds: Adjusted p-value: 0.05 and Log2 fold change: 1. Only one of the 11 genes is a immne-related gene of the list that I created manually.
So, you are saying you have a total of 11 DE genes of which 1 overlaps with a hand-picked list of genes of interest?
Exactly. And I would like to validate experimentally this analysis later, so I think if I can get more genes would be better.
That doesn't sound as if there's a lot to be done here. 11 DEG is not a whole lot to begin with, which makes me think that these samples aren't really that different (or there's a lot of variability between the replicates). You can always check the p-value (not the adjusted p-value) of the genes of interest, which will give you some insights into whether these immuno-related genes show any promise in these samples. After all, pre-filtering mostly influences the severity with which the "raw" p-values are adjusted as Ian has pointed out in his answer below.
Thank you so much for your help