Hello, I am analyzing total RNA-seq results and checking them in IGV.
I have noticed that the results from our experimental samples are different from the public data downloaded from GEO. The top 6 tracks are from our experimental samples, and all the tracks below the reference are from the public data.
What I am curious about is why there are small read counts between the exons in our data, but in the public data, which used the same cell line and the same drug treatment, there are none and the results are clean. At which stage do these differences arise? Should we consider the small read counts in our samples as noise?
1 answer
Intronic reads as seen here are pretty normal in total RNA-seq, as you're capturing more nascent transcripts than you would after polyA selection. Sometimes, these reads can be a pretty significant fraction of total reads, especially in certain tissues (e.g. brain) and developmental context (tend to be more in fetal than adult).
Is the data in GEO also total RNA-seq or is it polyA selected? It looks like sequencing depth may also just be lower. Generally, we do not throw out intronic reads. This wouldn't concern me too much, my guess is the differences stem from library prep method or sequencing depth.
Log in to answer this question.
-
Exclude from analyzing rna-seq samples
written by dropswater •
-
ONT methylation data background noise
written by marco.barr •
-
Differential gene expression pattern changes while using nuclear/unspliced counts
written by akilabioinfo •
-
Job: Senior Data Scientist - Hit Discovery at AstraZeneca (Cambridge, UK)
written by Sergio Martínez Cuesta
-
Job: Full-Stack-Developer with knowledge of managing biological data
written by D4office •
-
Determining RNAseq background noise
written by marcos.georgiades.18 •
-
Tool: [Crushing the bureaucracy in bioinformatics] Introducing Skymap-JupyterHub: accessing and uti…
written by btsui
-
DiffBind - normalization methods with biologically-relevant differences in signal-to-noise
written by reskejak •
-
Data sets suitable for analysing gene expression
written by fawazfebin
-
Does normalization/pooling in library prep affect read counts after sequencing?
written by snamjoshi87 •
Assuming all display parameters are equal.
It also looks like all peaks across the whole length of the image are higher in your dataset. Have you compared overall sequencing depth between the local/public data sets?
No, not yet.. I'll check that out. Thank you!!