Blood contamination in RNA-seq analysis

Question

[Bastien Hervé]

Hi,

We use mainstream pipeline (STAR -> FeatureCounts -> Deseq2 -> Heatmap) to set up a differential expression analysis on our RNAseq data. According to our results and discussions with biologists, a blood contamination has been detected... We have high expressed Hba that distort the analysis.

Question is : where should we remove blood contamination in our analysis ? After FeatureCounts or just before the heatmap printing ? Thanks for advises

[GenoMax]

Trying to imagine what could have happened here. What kind of sample was it originally supposed to be? If a sample has been contaminated then there may remain lingering questions about any results you get. Unless you have an irreplaceable sample and have to work with what you have.

Answer 1

[Jean-Karim Heriche]

To rephrase @genomax' comment, if the starting material is contaminated, any downstream analysis will be suspect. Unless you can restrict your analysis to sequences that are guaranteed not to be present in blood, I don't see how you could clean up the data and remove any doubt about the final results.
Find out how contamination occurs, take steps to prevent it then process a new sample.