I want to compare the gene expression of similar tissue-type coming from one SE and another PE dataset (each having 3 biological replicates in both). And also both are from different labs with different protocols.
So I tried to normalize both the datasets with normTransform in DESeq2 package from the raw counts but no help.
Any solution how can normalize both the datasets to compare gene expression directly.
I know this question must have been asked before. I tried my best to find some similar thread, but couldn't get a proper one. Please share a link if you have come across before.
•
written
by
sangram_keshari
2 answers
I don't see why there's an issue at the bioinformatics level (does that comparison makes sense, I cannot tell). Do you get any error message?
For example EdgeR is only taking raw counts as input, obtained for example with FeatureCounts, independently of the protocol used and the normalisation should then not be an issue.
You can refer to this nice paper if you want to get Differentially Expressed Genes (DEG): https://f1000research.com/articles/5-1408/v3
I also don't think this comparison is a good protocol. If you are trying to ask the question whether there is a difference between SE and PE expression data that would be possible, EXCEPT they are different samples, different labs, and different protocols.
How do you know regardless of whatever computational pipeline/protocol you do, whether it is SE vs PE or the other issues?
This paper can give some insight into the expected differences and results could be:(https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4531809/)
As well as other discussions:
Paired end vs Single end Differential gene expression analysis
Log in to answer this question.
-
Guidance Needed: Best Practices for Handling Technical Replicates in RNA-seq Analysis
written by hellothere •
-
Can I analyze single-end and paired-end RNA-Seq data together?
written by roxana •
-
RNA seq
written by Byonkesh •
-
Recommended way to normalize SmartSeq2 gene expression matrix to better match 10X expression data
written by Cookin •
-
Combining bulk RNA-Seq datasets from different studies with no common groups across them
written by dolphin-sea •
-
Compare gene expression between scRNAseq and bulk RNAseq data
written by ngs101.com
-
Best way to handle samples with both single end and paired end?
written by ichbinlynn93 •
-
Compare Microarray with RNA-Seq
written by sangram_keshari
-
Meta data analysis
written by rrsowmya •
-
FPKM values from SE vs. PE RNA-Seq libraries
written by s_herrera •
What does
but no helpmean?After normalization also, The scaling in both the datasets is different, which is not comparable.
Note: I want to compare relative expression.
Is this based on an MAplot or how do you see that the scales are different?
Here is a PCA plot: https://imgur.com/a/sv6zEw8
Towards extreme left the PE datasets and towards extreme right SE.
You can add
batchvariable to DESeq2 design while creating theDESeq object for the first time. Check the following linksAlso check PCA (after rlog/vst normalization with batch) by coloring SE & PE. If SE and PE samples belong to two different biological conditions/groups, then there is noway you can tell whether the difference is due to tech or biology.