Why and how to discard mapped reads based on strandedness?

Question

[Beginner]

Hi,

how can someone filter out reads based on the library strandedness? In a Ribo-Seq analysis, the Ribo-Seq reads mapping to the second-strand and for the RNA-Seq dataset reads mapping to the first-strand were discarded.

Why would someone do this and how to do it (aligner was STAR).

Thank you for your help!

Answer 1

[swbarnes2]

I'm not sure most people filter at the bam level, but when you use a program to assign reads to genes, you can tell those programs that your data is stranded, and that it should only count reads aligning in the correct orientation.

STAR can be instructed to output the gene counts using all three assumptions ( unstranded, forward, reverse), and then you pick the column of counts that matches what you did. Or, programs like featureCounts and RSEM can be told what strandedness to apply.

[Beginner]

Do you know why someone would discard those reads? Is the reason why they choose to discard Ribo-Seq reads mapping to the second-strand and for the RNA-Seq dataset reads mapping to the first-strand because the "normal non-anti-sense" read of RNA-Seq should be reverse? And a normal Ribo-Seq read will be anti-sense to the RNA-seq mapping orientation (forward)?