Is the read counting strand-specific?

[Mjaraespejo]

Hello,

I have used TPMCalculator to quantify RNA expression using bam files obtained using HISAT2.

For some genes I am getting extremely high intronic TPM counts when compared to exon values in the same gene. I have inspected those genes and I found that some introns overlap other genes in the opposite strand, partially or completely (see attached picture). So, when the TPM is calculated, the intron siganl seems to be inflated and doesn't correlate with the exonic values which are very low or zero.

To run TPMCalculator I use this commans:
TPMCalculator -v -g $annotation -k gene_id -t transcript_id -p -b $bam

My question is, when TPM counts the reads per feature, is the counting strand-specific?

Thank you,
Manuel.

[r78v10a07]

TPMCalculator is not strand specific. This is a feature we will implement for the next release.
In your particular case, TPMCalculator can not determine whether a reads belong to the exon or intro of two overlapping genes.
In this case, you could use the *.uni files which exclude reads on overlapping features of each gene.

[Mjaraespejo]

Hi Roberto,

Thanks for your reply. I'll try to use the .uni TPM values.

Manuel.