I'm looking at RNAseq data from CCLE. The data is paired-end.

Take the cell line Hs578T and the gene HRAS as an example.

The cell line carries a G12D mutation (c.35G>A), so the change in cds is:

ggc ggtgtgggca agagtgcgct g - Wildtype CDS
gAc ggtgtgggca agagtgcgct g - Mutant CDS

My question is, when I grep the mutant CDS gAcggtgtgggcaagagtgcgctg, I do not get a match in my .bam file. But when I grep the reverse complement (i.e., anti-sense sequence), I get matches coming from both mates.

My question is, why are there reads in the .bam file that correspond to the anti-sense strand of the DNA? Shouldn't all the reads of mRNA be the CDS?

I got the answer - The punch line is that the sequence in .bam files is ALWAYS '+' strand of the reference, no matter what mate the read came from, or which strand the gene sits on.


1 Answer 1


Yes, the mRNA results should include CDSs (not only, don't forget you also have UTR sequences which are non-coding so are not part of the CDS), but there is no reason why they should all be from genes on the + strand of the genome. Remember that the distinction between "sense/+" and "antisense/-" is essentially a human invention; it doesn't imply that the cell considers one strand to be in any way superior than the other, and genes are pretty randomly distributed across both strands.

This means that when you align to a human reference genome, you will always have some sequences aligning to the + strand, if the RNAs were transcribed from genes that happen to be on the sense strand, and others will align to the - strand, if they were transcribed from genes found in the antisense strand.

This is completely normal and expected, there is nothing wrong with your data. For example, human tP53 is on the antisense strand (note how the arrows are all pointing to the left):

screnshot of the UCSC genome browser showing tp53

  • $\begingroup$ Can you help me understand this: does the bam file sequence always correspond to the +strand of the reference genome, regardless of which mate it came from? $\endgroup$
    – geom_na
    Apr 25, 2023 at 15:37
  • 1
    $\begingroup$ @Nomad420 yes, exactly. Basically, we arbitrarily decided that "this is the 'sense' strand" and then all coordinates are given with respect to that strand. $\endgroup$
    – terdon
    Apr 25, 2023 at 15:49
  • $\begingroup$ Thank you. This clears up a lot of my confusion! $\endgroup$
    – geom_na
    Apr 25, 2023 at 16:06
  • $\begingroup$ Sequences in a bam ar always in the forward direction, with respect to the reference, even if the original read was in the opposite orientation. The flag indicates if the read as you see it in the bam is just like the original read, or if the bam is displaying the rev-comp. $\endgroup$
    – swbarnes2
    Apr 25, 2023 at 20:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.