Reads partly within an exon

Question

[benjyrolls]

Hi

am doing RNA Seq analysis of my data sets and came out with these stats. link attached. I have reads within an exon and some partly within an exon as well as some which are within an intron and some which are fullyintergenic. What do these mean ?

[GenoMax]

Any chance of genomic DNA contamination in your data?

[benjyrolls]

possibly or unspliced mRNA in the sample. ?

[GenoMax]

Have you checked to see the actual distribution? Is it uniform across the genome?

[benjyrolls]

Have you checked to see the actual distribution? Is it uniform across the genome?

am not sure if i understand what you mean genomax.

[GenoMax]

Are those problem reads located across multiple areas of the genome or only some parts?

[benjyrolls]

Its a quantification summary of all the reads for all the samples and their replicates

[GenoMax]

I got that. But have you checked the alignments in a genome browser.

[benjyrolls]

yes I have is there a portion I could check to find the coverage and figure whats going on. browser view

Assuming these are genomic DNA contamination, should this a cause for concern compared with the number of reads 84% that are within an exon ?

[microfuge]

intron retention is not that uncommon in mammals ( http://www.biorxiv.org/content/biorxiv/early/2017/05/09/135624.full.pdf ). In plants quite common in plants (https://www.ncbi.nlm.nih.gov/pubmed/15341630).

[WouterDeCoster]

I also suspect DNA contamination, which is pretty damn annoying.

One concern is: do the levels of contamination differ between your biological groups? If so you have a pretty big batch effect. If it's all the same across your samples then you have a lot of noise in your data.

But we don't know what the aim of your analysis is.

[benjyrolls]

we don't know what the aim of your analysis is.

The aim of the analysis is to find the effect of SOX2 knockout in the stomach

[benjyrolls]

If its genomic DNA contamination what are the effects on the experimental outcome?

[GenoMax]

Your expression counts would be off which may affect your DE results. You can do DE analysis and see if genes showing up as DE make sense (and do not have reads of the kind you are describing above for those genes). Consider choosing a read count methodology that ignores such reads. As @Wouter pointed out we don't know if there is a batch effect (i.e. some samples are more affected than others).

[benjyrolls]

Consider choosing a read count methodology that ignores such reads

Am using Cuffdiff am not sure if this can do this ?

we don't know if there is a batch effect (i.e. some samples are more affected than others).

I need to do further analysis to see this.

You can do DE analysis and see if genes showing up as DE make sense (and do not have reads of the kind you are describing above for those genes).

I have a corresponding ChIP-Seq experiment which is targeting SOX2 genes. I have found these genes known to be targeted by Sox2( for example genes: Usp37 and Pnkd ) and these two genes are also present in the RNA-Seq experiment. However in low quantities, in the SOX2 KO samples ( from comparing expression plots of the KO and WT samples) . Based on these known genes from the ChIP-Seq experiment can I draw these conclusions that these two genes are in low quantities in the knockout samples ?

[benjyrolls]

if I did pathway analysis for genes targeted by SOX2 and found those genes also present in RNA-Seq experiment is this also right ?